conference
stringclasses 6
values | title
stringlengths 2
176
| abstract
stringlengths 2
5k
| decision
stringclasses 11
values |
|---|---|---|---|
ICLR.cc/2021/Conference | Improving Self-supervised Pre-training via a Fully-Explored Masked Language Model | Masked Language Model (MLM) framework has been widely adopted for self-supervised language pre-training. In this paper, we argue that randomly sampled masks in MLM would lead to undesirably large gradient variance. Thus, we theoretically quantify the gradient variance via correlating the gradient covariance with the Hamming distance between two different masks (given a certain text sequence). To reduce the variance due to the sampling of masks, we propose a fully-explored masking strategy, where a text sequence is divided into a certain number of non-overlapping segments. Thereafter, the tokens within one segment are masked for training. We prove, from a theoretical perspective, that the gradients derived from this new masking schema have a smaller variance and can lead to more efficient self-supervised training. We conduct extensive experiments on both continual pre-training and general pre-training from scratch. Empirical results confirm that this new masking strategy can consistently outperform standard random masking. Detailed efficiency analysis and ablation studies further validate the advantages of our fully-explored masking strategy under the MLM framework. | Reject |
ICLR.cc/2019/Conference | Clean-Label Backdoor Attacks | Deep neural networks have been recently demonstrated to be vulnerable to backdoor attacks. Specifically, by altering a small set of training examples, an adversary is able to install a backdoor that can be used during inference to fully control the model’s behavior. While the attack is very powerful, it crucially relies on the adversary being able to introduce arbitrary, often clearly mislabeled, inputs to the training set and can thus be detected even by fairly rudimentary data filtering. In this paper, we introduce a new approach to executing backdoor attacks, utilizing adversarial examples and GAN-generated data. The key feature is that the resulting poisoned inputs appear to be consistent with their label and thus seem benign even upon human inspection. | Reject |
ICLR.cc/2019/Conference | CGNF: Conditional Graph Neural Fields | Graph convolutional networks have achieved tremendous success in the tasks of graph node classification. These models could learn a better node representation through encoding the graph structure and node features. However, the correlation between the node labels are not considered. In this paper, we propose a novel architecture for graph node classification, named conditional graph neural fields (CGNF). By integrating the conditional random fields (CRF) in the graph convolutional networks, we explicitly model a joint probability of the entire set of node labels, thus taking advantage of neighborhood label information in the node label prediction task.
Our model could have both the representation capacity of graph neural networks and the prediction power of CRFs. Experiments on several graph datasets demonstrate effectiveness of CGNF. | Reject |
ICLR.cc/2020/Conference | Pre-trained Contextual Embedding of Source Code | The source code of a program not only serves as a formal description of an executable task, but it also serves to communicate developer intent in a human-readable form. To facilitate this, developers use meaningful identifier names and natural-language documentation. This makes it possible to successfully apply sequence-modeling approaches, shown to be effective in natural-language processing, to source code. A major advancement in natural-language understanding has been the use of pre-trained token embeddings; BERT and other works have further shown that pre-trained contextual embeddings can be extremely powerful and can be finetuned effectively for a variety of downstream supervised tasks. Inspired by these developments, we present the first attempt to replicate this success on source code. We curate a massive corpus of Python programs from GitHub to pre-train a BERT model, which we call Code Understanding BERT (CuBERT). We also pre-train Word2Vec embeddings on the same dataset. We create a benchmark of five classification tasks and compare finetuned CuBERT against sequence models trained with and without the Word2Vec embeddings. Our results show that CuBERT outperforms the baseline methods by a margin of 2.9-22%. We also show its superiority when finetuned with smaller datasets, and over fewer epochs. | Reject |
ICLR.cc/2021/Conference | Federated learning using mixture of experts | Federated learning has received attention for its efficiency and privacy benefits,in settings where data is distributed among devices. Although federated learning shows significant promise as a key approach when data cannot be shared or centralized, current incarnations show limited privacy properties and have short-comings when applied to common real-world scenarios. One such scenario is heterogeneous data among devices, where data may come from different generating distributions. In this paper, we propose a federated learning framework using a mixture of experts to balance the specialist nature of a locally trained model with the generalist knowledge of a global model in a federated learning setting. Our results show that the mixture of experts model is better suited as a personalized model for devices when data is heterogeneous, outperforming both global and lo-cal models. Furthermore, our framework gives strict privacy guarantees, which allows clients to select parts of their data that may be excluded from the federation. The evaluation shows that the proposed solution is robust to the setting where some users require a strict privacy setting and do not disclose their models to a central server at all, opting out from the federation partially or entirely. The proposed framework is general enough to include any kind of machine learning models, and can even use combinations of different kinds. | Reject |
ICLR.cc/2023/Conference | Open-Vocabulary Semantic Segmentation with Mask-adapted CLIP | Open-vocabulary semantic segmentation aims to segment an image into semantic regions according to text descriptions, which may not have been seen during training. Recent two-stage methods first generate class-agnostic mask proposals and then leverage pre-trained vision-language models, e.g., CLIP, to classify masked regions. We identify the performance bottleneck of this paradigm to be the pre-trained CLIP model, since it does not perform well on masked images. To address this, we propose to finetune CLIP on a collection of masked image regions and their corresponding text descriptions. We collect training data by mining an existing image-caption dataset (e.g., COCO Captions), using CLIP to match masked image regions to nouns in the image captions. Compared with the more precise and manually annotated segmentation labels with fixed classes (e.g., COCO-Stuff), we find our noisy but diverse dataset can better retain CLIP's generalization ability. Along with finetuning the entire model, we utilize the "blank" areas in masked images using a method we dub mask prompt tuning. Experiments demonstrate mask prompt tuning brings significant improvement without modifying any weights of CLIP, and it can further improve a fully finetuned model. In particular, when trained on COCO and evaluated on ADE20K-150, our best model achieves 29.6% mIoU, which is +8.5% higher than the previous state-of-the-art. For the first time, open-vocabulary generalist models match the performance of supervised specialist models in 2017 without dataset-specific adaptations. | Withdrawn |
ICLR.cc/2023/Conference | Dual Ensembled Multiagent Q-Learning with Hypernet Regularizer | Overestimation in the temporal-difference single-agent reinforcement learning has been widely studied, where the variance in value estimation causes overestimation of the maximal target value due to Jensen's inequality. Instead, overestimation in multiagent settings has received little attention though it can be even more severe. One kind of pioneer work extends ensemble methods from single-agent deep reinforcement learning to address the multiagent overestimation by discarding the large target values among the ensemble. However, its ability is limited by the ensemble diversity. Another kind of work softens the maximum operator in the Bellman equation to avoid large target values, but also leads to sub-optimal value functions. Unlike previous works, in this paper, we address the multiagent overestimation by analyzing its underlying causes in an estimation-optimization iteration manner. We show that the overestimation in multiagent value-mixing Q-learning not only comes from the overestimation of target Q-values but also accumulates in the online Q-network's optimization step. Therefore, first, we integrate the random ensemble and in-target minimization into the estimation of target Q-values to derive a lower update target. Second, we propose a novel hypernet regularizer on the learnable terms of the online global Q-network to further reduce overestimation. Experiments on various kinds of tasks demonstrate that the proposed method consistently addresses the overestimation problem while previous works fail. | Reject |
ICLR.cc/2023/Conference | Retrieval-based Controllable Molecule Generation | Generating new molecules with specified chemical and biological properties via generative models has emerged as a promising direction for drug discovery. However, existing methods require extensive training/fine-tuning with a large dataset, often unavailable in real-world generation tasks. In this work, we propose a new retrieval-based framework for controllable molecule generation. We use a small set of exemplar molecules, i.e., those that (partially) satisfy the design criteria, to steer the pre-trained generative model towards synthesizing molecules that satisfy the given design criteria. We design a retrieval mechanism that retrieves and fuses the exemplar molecules with the input molecule, which is trained by a new self-supervised objective that predicts the nearest neighbor of the input molecule. We also propose an iterative refinement process to dynamically update the generated molecules and retrieval database for better generalization. Our approach is agnostic to the choice of generative models and requires no task-specific fine-tuning. On various tasks ranging from simple design criteria to a challenging real-world scenario for designing lead compounds that bind to the SARS-CoV-2 main protease, we demonstrate our approach extrapolates well beyond the retrieval database, and achieves better performance and wider applicability than previous methods. | Accept: notable-top-25% |
ICLR.cc/2020/Conference | New Loss Functions for Fast Maximum Inner Product Search | Quantization based methods are popular for solving large scale maximum inner product search problems. However, in most traditional quantization works, the objective is to minimize the reconstruction error for datapoints to be searched. In this work, we focus directly on minimizing error in inner product approximation and derive a new class of quantization loss functions. One key aspect of the new loss functions is that we weight the error term based on the value of the inner product, giving more importance to pairs of queries and datapoints whose inner products are high. We provide theoretical grounding to the new quantization loss function, which is simple, intuitive and able to work with a variety of quantization techniques, including binary quantization and product quantization. We conduct experiments on public benchmarking datasets \url{http://ann-benchmarks.com} to demonstrate that our method using the new objective outperforms other state-of-the-art methods. We are committed to release our source code. | Reject |
ICLR.cc/2023/Conference | Faster Neural Architecture "Search" for Deep Image Prior | Deep image prior (DIP) is known for leveraging the spectral bias of the convolutional neural network (CNN) towards lower frequencies in various single-image restoration tasks. Such inductive bias has been widely attributed to the network architecture. Existing studies therefore either handcraft the architecture or use automated neural architecture search (NAS). However, there is still a lack of understanding on how the architectural choice corresponds to the image to be restored, leading to an excessively large search space that is both time and computationally-expensive for typical NAS techniques. As a result, the architecture is often searched and fixed for the whole dataset, while the best-performing one could be image-dependent. Moreover, common architecture search requires ground truth supervision, which is often not accessible. In this work, we present a simple yet effective \emph{training-free} approach to estimate the required architecture for \emph{every image} in advance. This is motivated by our empirical findings that the width and depth of a good network prior are correlated with the texture of the image, which can be estimated during pre-processings. Accordingly, the design space is substantially shrunk to a handful of subnetworks within a given large network. The experiments on denoising across different noise levels show that a subnetwork with proper setups could be a more effective network prior than the original network while being highly under-parameterized, making it not critically require early-stopping as with the original large network. | Withdrawn |
ICLR.cc/2023/Conference | Text-Conditioned Graph Generation Using Discrete Graph Variational Autoencoders | Inspired by recent progress in text-conditioned image generation, we propose a model for the novel problem of text-conditioned graph generation. In this paper we introduce the Vector Quantized Text To Graph generator (VQ-T2G), a discrete graph variational autoencoder and autoregressive transformer for generating general graphs conditioned on text. We curate two multimodal datasets of graphs paired with text, a real-world dataset of 8000 subgraphs from the Wikipedia link network and a dataset of over 5000 synthetic graphs. Experimental results on these datasets demonstrate that VQ-T2G synthesises novel graphs with structure aligned with the text conditioning. Additional experiments in the unconditioned graph generation setting show VQ-T2G is competitive with existing unconditioned graph generation methods across various graph metrics. Code will be released on github following paper acceptance. | Withdrawn |
ICLR.cc/2023/Conference | FixEval: Execution-based Evaluation of Program Fixes for Competitive Programming Problems | The increasing complexity of software has led to a drastic rise in time and costs for identifying and fixing bugs. Various approaches are explored in the literature to generate fixes for buggy code automatically. However, due to the large combinatorial space of possible fixes for a particular bug, few tools and datasets are available to evaluate model generated fixes effectively. In this work, we introduce FixEval, a benchmark comprising buggy code submissions to competitive programming problems and their respective fixes.
FixEval is composed of a rich test suite to evaluate and assess the correctness of model-generated program fixes and further information regarding time and memory constraints and acceptance based on a verdict. We consider two Transformer language models pretrained on programming languages as our baselines and compare them using match-based and execution-based evaluation metrics. Our experiments show that match-based metrics do not reflect model-generated program fixes accurately. At the same time, execution-based methods evaluate programs through all cases and scenarios designed explicitly for that solution.
Therefore, we believe FixEval provides a step towards real-world automatic bug fixing and model-generated code evaluation. The dataset and models are open-sourced.\footnote{\url{https://github.com/FixEval/FixEval_official}} | Withdrawn |
ICLR.cc/2021/Conference | Lipschitz Recurrent Neural Networks | Viewing recurrent neural networks (RNNs) as continuous-time dynamical systems, we propose a recurrent unit that describes the hidden state's evolution with two parts: a well-understood linear component plus a Lipschitz nonlinearity. This particular functional form facilitates stability analysis of the long-term behavior of the recurrent unit using tools from nonlinear systems theory. In turn, this enables architectural design decisions before experimentation. Sufficient conditions for global stability of the recurrent unit are obtained, motivating a novel scheme for constructing hidden-to-hidden matrices. Our experiments demonstrate that the Lipschitz RNN can outperform existing recurrent units on a range of benchmark tasks, including computer vision, language modeling and speech prediction tasks. Finally, through Hessian-based analysis we demonstrate that our Lipschitz recurrent unit is more robust with respect to input and parameter perturbations as compared to other continuous-time RNNs. | Accept (Poster) |
ICLR.cc/2023/Conference | ETSformer: Exponential Smoothing Transformers for Time-series Forecasting | Transformers have recently been actively studied for time-series forecasting. While often showing promising results in various scenarios, traditional Transformers are not designed to fully exploit the characteristics of time-series data and thus suffer some fundamental limitations, e.g., they are generally not decomposable or interpretable, and are neither effective nor efficient for long-term forecasting. In this paper, we propose ETSformer, a novel time-series Transformer architecture, which exploits the principle of exponential smoothing methods in improving Transformers for time-series forecasting. Specifically, ETSformer leverages a novel level-growth-seasonality decomposed Transformer architecture which leads to more interpretable and disentangled decomposed forecasts. We further propose two novel attention mechanisms -- the exponential smoothing attention and frequency attention, which are specially designed to overcome the limitations of the vanilla attention mechanism for time-series data. Extensive experiments on various time-series benchmarks validate the efficacy and advantages of the proposed method. Code is attached in the supplementary material, and will be made publicly available. | Reject |
ICLR.cc/2022/Conference | Stein Latent Optimization for Generative Adversarial Networks | Generative adversarial networks (GANs) with clustered latent spaces can perform conditional generation in a completely unsupervised manner. In the real world, the salient attributes of unlabeled data can be imbalanced. However, most of existing unsupervised conditional GANs cannot cluster attributes of these data in their latent spaces properly because they assume uniform distributions of the attributes. To address this problem, we theoretically derive Stein latent optimization that provides reparameterizable gradient estimations of the latent distribution parameters assuming a Gaussian mixture prior in a continuous latent space. Structurally, we introduce an encoder network and novel unsupervised conditional contrastive loss to ensure that data generated from a single mixture component represent a single attribute. We confirm that the proposed method, named Stein Latent Optimization for GANs (SLOGAN), successfully learns balanced or imbalanced attributes and achieves state-of-the-art unsupervised conditional generation performance even in the absence of attribute information (e.g., the imbalance ratio). Moreover, we demonstrate that the attributes to be learned can be manipulated using a small amount of probe data. | Accept (Poster) |
ICLR.cc/2023/Conference | Revisiting Activation Function Design for Improving Adversarial Robustness at Scale | Modern ConvNets typically use ReLU activation function. Recently smooth activation functions have been used to improve their accuracy. Here we study the role of smooth activation function from the perspective of adversarial robustness. We find that ReLU activation function significantly weakens adversarial training due to its non-smooth nature. Replacing ReLU with its smooth alternatives allows adversarial training to find harder adversarial training examples and to compute better gradient updates for network optimization.
We focus our study on the large-scale ImageNet dataset. On ResNet-50, switching from ReLU to the smooth activation function SILU improves adversarial robustness from 33.0% to 42.3%, while also improving accuracy by 0.9% on ImageNet. Smooth activation functions also scale well with larger networks: it helps EfficientNet-L1 to achieve 82.2% accuracy and 58.6% robustness, largely outperforming the previous state-of-the-art defense by 9.5% for accuracy and 11.6% for robustness. Models are available at https://rb.gy/qt8jya. | Reject |
ICLR.cc/2021/Conference | HyperGrid Transformers: Towards A Single Model for Multiple Tasks | Achieving state-of-the-art performance on natural language understanding tasks typically relies on fine-tuning a fresh model for every task. Consequently, this approach leads to a higher overall parameter cost, along with higher technical maintenance for serving multiple models. Learning a single multi-task model that is able to do well for all the tasks has been a challenging and yet attractive proposition. In this paper, we propose HyperGrid Transformers, a new Transformer architecture that leverages task-conditioned hyper networks for controlling its feed-forward layers. Specifically, we propose a decomposable hypernetwork that learns grid-wise projections that help to specialize regions in weight matrices for different tasks. In order to construct the proposed hypernetwork, our method learns the interactions and composition between a global (task-agnostic) state and a local task-specific state. We conduct an extensive set of experiments on GLUE/SuperGLUE. On the SuperGLUE test set, we match the performance of the state-of-the-art while being $16$ times more parameter efficient. Our method helps bridge the gap between fine-tuning and multi-task learning approaches. | Accept (Poster) |
ICLR.cc/2023/Conference | Improving the Strength of Human-Like Models in Chess | Designing AI systems that capture human-like behavior has attracted growing attention in applications where humans may want to learn from, or need to collaborate with, these AI systems. Many existing works in designing human-like AI have taken a supervised learning approach that learns from data of human behavior, with the goal of creating models that can accurately predict human behavior. While this approach has shown success in capturing human behavior at different skill levels and even identifying individual behavioral styles, it also suffers from the drawback of mimicking human mistakes. Moreover, existing models only capture a snapshot of human behavior, leaving the question of how to improve them---e.g., from one human skill level to a stronger one---largely unanswered. Using chess as an experimental domain, we investigate the question of teaching an existing human-like model to be stronger using a data-efficient curriculum, while maintaining the model's human similarity. To achieve this goal, we extend the concept of curriculum learning to settings with multiple labeling strategies, allowing us to vary both the curriculum (dataset) and the teacher (labeling strategy). We find that the choice of teacher has a strong impact on both playing strength and human similarity; for example, a teacher that is too strong can be less effective at improving playing strength and degrade human similarity more rapidly. We also find that the choice of curriculum can impact these metrics, but to a smaller extent; for example, training on a curriculum of human mistakes provides only a marginal benefit over training on a random curriculum. Finally, we show that our strengthened models achieve human similarity on datasets corresponding to their strengthened level of play, suggesting that our curriculum training methodology is improving them in human-like steps. | Reject |
ICLR.cc/2020/Conference | A Closer Look at the Optimization Landscapes of Generative Adversarial Networks | Generative adversarial networks have been very successful in generative modeling, however they remain relatively challenging to train compared to standard deep neural networks. In this paper, we propose new visualization techniques for the optimization landscapes of GANs that enable us to study the game vector field resulting from the concatenation of the gradient of both players. Using these visualization techniques we try to bridge the gap between theory and practice by showing empirically that the training of GANs exhibits significant rotations around LSSP, similar to the one predicted by theory on toy examples. Moreover, we provide empirical evidence that GAN training seems to converge to a stable stationary point which is a saddle point for the generator loss, not a minimum, while still achieving excellent performance. | Accept (Poster) |
ICLR.cc/2022/Conference | Dual Lottery Ticket Hypothesis | Fully exploiting the learning capacity of neural networks requires overparameterized dense networks. On the other side, directly training sparse neural networks typically results in unsatisfactory performance. Lottery Ticket Hypothesis (LTH) provides a novel view to investigate sparse network training and maintain its capacity. Concretely, it claims there exist winning tickets from a randomly initialized network found by iterative magnitude pruning and preserving promising trainability (or we say being in trainable condition). In this work, we regard the winning ticket from LTH as the subnetwork which is in trainable condition and its performance as our benchmark, then go from a complementary direction to articulate the Dual Lottery Ticket Hypothesis (DLTH): Randomly selected subnetworks from a randomly initialized dense network can be transformed into a trainable condition and achieve admirable performance compared with LTH --- random tickets in a given lottery pool can be transformed into winning tickets. Specifically, by using uniform-randomly selected subnetworks to represent the general cases, we propose a simple sparse network training strategy, Random Sparse Network Transformation (RST), to substantiate our DLTH. Concretely, we introduce a regularization term to borrow learning capacity and realize information extrusion from the weights which will be masked. After finishing the transformation for the randomly selected subnetworks, we conduct the regular finetuning to evaluate the model using fair comparisons with LTH and other strong baselines. Extensive experiments on several public datasets and comparisons with competitive approaches validate our DLTH as well as the effectiveness of the proposed model RST. Our work is expected to pave a way for inspiring new research directions of sparse network training in the future. Our code is available at https://github.com/yueb17/DLTH. | Accept (Poster) |
ICLR.cc/2023/Conference | Comparing Auxiliary Tasks for Learning Representations for Reinforcement Learning | Learning state representations has gained steady popularity in reinforcement learning (RL) due to its potential to improve both sample efficiency and returns on many environments. A straightforward and efficient method is to generate representations with a distinct neural network trained on an auxiliary task, i.e. a task that differs from the actual RL task. While a whole range of such auxiliary tasks has been proposed in the literature, a comparison on typical continuous control benchmark environments is computationally expensive and has, to the best of our knowledge, not been performed before. This paper presents such a comparison of common auxiliary tasks, based on hundreds of agents trained with state-of-the-art off-policy RL algorithms. We compare possible improvements in both sample efficiency and returns for environments ranging from simple pendulum to a complex simulated robotics task. Our findings show that representation learning with auxiliary tasks is beneficial for environments of higher dimension and complexity, and that learning environment dynamics is preferable to predicting rewards. We believe these insights will enable other researchers to make more informed decisions on how to utilize representation learning for their specific problem. | Withdrawn |
ICLR.cc/2023/Conference | A Learning Based Hypothesis Test for Harmful Covariate Shift | The ability to quickly and accurately identify covariate shift at test time is a critical and often overlooked component of safe machine learning systems deployed in high-risk domains. While methods exist for detecting when predictions should not be made on out-of-distribution test examples, identifying distributional level differences between training and test time can help determine when a model should be removed from the deployment setting and retrained. In this work, we define harmful covariate shift (HCS) as a change in distribution that may weaken the generalization of a predictive model. To detect HCS, we use the discordance between an ensemble of classifiers trained to agree on training data and disagree on test data. We derive a loss function for training this ensemble and show that the disagreement rate and entropy represent powerful discriminative statistics for HCS. Empirically, we demonstrate the ability of our method to detect harmful covariate shift with statistical certainty on a variety of high-dimensional datasets. Across numerous domains and modalities, we show state-of-the-art performance compared to existing methods, particularly when the number of observed test samples is small. | Accept: poster |
ICLR.cc/2023/Conference | Effective passive membership inference attacks in federated learning against overparameterized models | This work considers the challenge of performing membership inference attacks in a federated learning setting ---for image classification--- where an adversary can only observe the communication between the central node and a single client (a passive white-box attack). Passive attacks are one of the hardest-to-detect attacks, since they can be performed without modifying how the behavior of the central server or its clients, and assumes *no access to private data instances*. The key insight of our method is empirically observing that, near parameters that generalize well in test, the gradient of large overparameterized neural network models statistically behave like high-dimensional independent isotropic random vectors. Using this insight, we devise two attacks that are often little impacted by existing and proposed defenses. Finally, we validated the hypothesis that our attack depends on the overparametrization by showing that increasing the level of overparametrization (without changing the neural network architecture) positively correlates with our attack effectiveness. | Accept: poster |
ICLR.cc/2021/Conference | What Matters for On-Policy Deep Actor-Critic Methods? A Large-Scale Study | In recent years, reinforcement learning (RL) has been successfully applied to many different continuous control tasks. While RL algorithms are often conceptually simple, their state-of-the-art implementations take numerous low- and high-level design decisions that strongly affect the performance of the resulting agents. Those choices are usually not extensively discussed in the literature, leading to discrepancy between published descriptions of algorithms and their implementations. This makes it hard to attribute progress in RL and slows down overall progress [Engstrom'20]. As a step towards filling that gap, we implement >50 such ``"choices" in a unified on-policy deep actor-critic framework, allowing us to investigate their impact in a large-scale empirical study. We train over 250'000 agents in five continuous control environments of different complexity and provide insights and practical recommendations for the training of on-policy deep actor-critic RL agents. | Accept (Oral) |
ICLR.cc/2023/Conference | COFS: COntrollable Furniture layout Synthesis | Realistic, scalable, and controllable generation of furniture layouts is essential for many applications in virtual reality, augmented reality, game development and synthetic data generation. The most successful current methods tackle this problem as a sequence generation problem which imposes a specific ordering on the elements of the layout, making it hard to exert fine-grained control over the attributes of a generated scene. Existing methods provide control through object-level conditioning, or scene completion, where generation can be conditioned on an arbitrary subset of furniture objects. However, attribute-level conditioning, where generation can be conditioned on an arbitrary subset of object attributes, is not supported. We propose COFS, a method to generate furniture layouts that enables fine-grained control through attribute-level conditioning. For example, COFS allows specifying only the scale and type of objects that should be placed in the scene and the generator chooses their positions and orientations; or the position that should be occupied by objects can be specified and the generator chooses their type, scale, orientation, etc. Our results show both qualitatively and quantitatively that we significantly outperform existing methods on attribute-level conditioning. | Reject |
ICLR.cc/2021/Conference | NASOA: Towards Faster Task-oriented Online Fine-tuning | Fine-tuning from pre-trained ImageNet models has been a simple, effective, and popular approach for various computer vision tasks. The common practice of fine-tuning is to adopt a default hyperparameter setting with a fixed pre-trained model, while both of them are not optimized for specific tasks and time constraints. Moreover, in cloud computing or GPU clusters where the tasks arrive sequentially in a stream, faster online fine-tuning is a more desired and realistic strategy for saving money, energy consumption, and CO2 emission. In this paper, we propose a joint Neural Architecture Search and Online Adaption framework named NASOA towards a faster task-oriented fine-tuning upon the request of users. Specifically, NASOA first adopts an offline NAS to identify a group of training-efficient networks to form a pretrained model zoo. We propose a novel joint block and macro-level search space to enable a flexible and efficient search. Then, by estimating fine-tuning performance via an adaptive model by accumulating experience from the past tasks, an online schedule generator is proposed to pick up the most suitable model and generate a personalized training regime with respect to each desired task in a one-shot fashion. The resulting model zoo is more training efficient than SOTA NAS models, e.g. 6x faster than RegNetY-16GF, and 1.7x faster than EfficientNetB3. Experiments on multiple datasets also show that NASOA achieves much better fine-tuning results, i.e. improving around 2.1% accuracy than the best performance in RegNet series under various time constraints and tasks; 40x faster compared to the BOHB method. | Reject |
ICLR.cc/2023/Conference | From Images to Textual Prompts: Zero-shot VQA with Frozen Large Language Models | Large language models (LLMs) have demonstrated excellent zero-shot generalization to new tasks. However, effective utilization of LLMs for zero-shot visual question-answering (VQA) remains challenging, primarily due to the modality disconnection and task disconnection between LLM and VQA task. End-to-end training on vision and language data may bridge the disconnections, but is inflexible and computationally expensive. To address this issue, we propose \emph{Img2Prompt}, a plug-and-play module that provides the prompts that can bridge the aforementioned modality and task disconnections, so that LLMs can perform VQA tasks without end-to-end training. In order to provide such prompts, we further employ LLM-agnostic models to provide prompts that can describe image content and self-constructed question-answer pairs, which can effectively guide LLM to perform VQA tasks. Img2Prompt offers the following benefits: 1) It is LLM-agnostic and can work with any LLM to perform VQA. 2) It renders end-to-end training unnecessary and significantly reduces the cost of deploying LLM for VQA tasks. 3) It achieves comparable or better performance than methods relying on end-to-end training. On the challenging A-OKVQA dataset, our method outperforms some few-shot methods by as much as 20\%. | Withdrawn |
ICLR.cc/2021/Conference | MoPro: Webly Supervised Learning with Momentum Prototypes | We propose a webly-supervised representation learning method that does not suffer from the annotation unscalability of supervised learning, nor the computation unscalability of self-supervised learning. Most existing works on webly-supervised representation learning adopt a vanilla supervised learning method without accounting for the prevalent noise in the training data, whereas most prior methods in learning with label noise are less effective for real-world large-scale noisy data. We propose momentum prototypes (MoPro), a simple contrastive learning method that achieves online label noise correction, out-of-distribution sample removal, and representation learning. MoPro achieves state-of-the-art performance on WebVision, a weakly-labeled noisy dataset. MoPro also shows superior performance when the pretrained model is transferred to down-stream image classification and detection tasks. It outperforms the ImageNet supervised pretrained model by +10.5 on 1-shot classification on VOC, and outperforms the best self-supervised pretrained model by +17.3 when finetuned on 1% of ImageNet labeled samples. Furthermore, MoPro is more robust to distribution shifts. Code and pretrained models are available at https://github.com/salesforce/MoPro. | Accept (Poster) |
ICLR.cc/2021/Conference | Imagine That! Leveraging Emergent Affordances for 3D Tool Synthesis | In this paper we explore the richness of information captured by the latent space of a vision-based generative model. The model combines unsupervised generative learning with a task-based performance predictor to learn and to exploit task-relevant object affordances given visual observations from a reaching task, involving a scenario and a stick-like tool. While the learned embedding of the generative model captures factors of variation in 3D tool geometry (e.g. length, width, and shape), the performance predictor identifies sub-manifolds of the embedding that correlate with task success. Within a variety of scenarios, we demonstrate that traversing the latent space via backpropagation from the performance predictor allows us to imagine tools appropriate for the task at hand. Our results indicate that affordances – like the utility for reaching – are encoded along smooth trajectories in latent space. Accessing these emergent affordances by considering only high-level performance criteria (such as task success) enables an agent to manipulate tool geometries in a targeted and deliberate way. | Reject |
ICLR.cc/2023/Conference | Noise$^+$2Noise: Co-taught De-noising Autoencoders for Time-Series Data | We consider the task of learning to recover clean signals given only access to noisy data. Recent work in computer vision has addressed this problem in the context of images using denoising autoencoders (DAEs). However, to date DAEs for learning from noisy data have not been explored in the context of time-series data. DAEs for denoising images often rely on assumptions unlikely to hold in the context of time series, \textit{e.g.}, multiple noisy samples of the same example. Here, we adapt DAEs to cleaning time-series data with noisy samples only. To recover the clean target signal when only given access to noisy target data, we leverage a noise-free auxiliary time-series signal that is related to the target signal. In addition to leveraging the relationship between the target signal and auxiliary signal, we iteratively filter and learn from clean samples using an approach based on co-teaching. Applied to the task of recovering carbohydrate values for blood glucose management, our approach reduces noise (MSE) in patient-reported carbohydrates from 72$g^2$ (95\% CI: 54,93) to 18$g^2$ (13,25), outperforming the best baseline (MSE = 33$g^2$ (27,43)). We demonstrate strong time-series denoising performance, extending the applicability of DAEs to a previously under-explored setting. | Reject |
ICLR.cc/2023/Conference | Reinforced Sample Reweighting Policy for Semi-supervised Learning | Semi-supervised learning (SSL) has been shown to be an effective paradigm for learning with less labeled data. To improve the performance of SSL, existing methods build sample reweighting or thresholding strategies to handle the category bias or erroneous pseudo labels. However, most of these existing methods are based on the heuristic hand-crafted rules, which require laborious adjustment, and may lead to sub-optimal solutions that cannot improve the model performance to the greatest extent. Here, to the best of our knowledge, we pioneer to develop an automatic strategy that boosts the performance of SSL. We introduce an end-to-end sample reweighting policy for semi-supervised learning, with a delicately designed Markov Decision Process (MDP) framework. The MDP framework is constructed with an agent network, which is optimized in a reward-driven manner, and receives the carefully designed state and action representations for decision reference. We also design a memory paradigm for computation-efficient representation construction and MDP solving. We further introduce a "pretraining-boosting" two-stage MDP curriculum where the agent network is firstly pretrained and then optimized continuously in the deployment phase to catch up with the constantly updated classification network. Extensive experiments demonstrate that our method achieves state-of-the-art performance on multiple datasets, outperforming previous advanced approaches such as FixMatch. | Withdrawn |
ICLR.cc/2020/Conference | Universal Adversarial Attack Using Very Few Test Examples | Adversarial attacks such as Gradient-based attacks, Fast Gradient Sign Method (FGSM) by Goodfellow et al.(2015) and DeepFool by Moosavi-Dezfooli et al. (2016) are input-dependent, small pixel-wise perturbations of images which fool state of the art neural networks into misclassifying images but are unlikely to fool any human. On the other hand a universal adversarial attack is an input-agnostic perturbation. The same perturbation is applied to all inputs and yet the neural network is fooled on a large fraction of the inputs. In this paper, we show that multiple known input-dependent pixel-wise perturbations share a common spectral property. Using this spectral property, we show that the top singular vector of input-dependent adversarial attack directions can be used as a very simple universal adversarial attack on neural networks. We evaluate the error rates and fooling rates of three universal attacks, SVD-Gradient, SVD-DeepFool and SVD-FGSM, on state of the art neural networks. We show that these universal attack vectors can be computed using a small sample of test inputs. We establish our results both theoretically and empirically. On VGG19 and VGG16, the fooling rate of SVD-DeepFool and SVD-Gradient perturbations constructed from observing less than 0.2% of the validation set of ImageNet is as good as the universal attack of Moosavi-Dezfooli et al. (2017a). To prove our theoretical results, we use matrix concentration inequalities and spectral perturbation bounds. For completeness, we also discuss another recent approach to universal adversarial perturbations based on (p, q)-singular vectors, proposed independently by Khrulkov & Oseledets (2018), and point out the simplicity and efficiency of our universal attack as the key difference. | Reject |
ICLR.cc/2021/Conference | Quantifying Task Complexity Through Generalized Information Measures | How can we measure the “complexity” of a learning task so that we can compare one task to another? From classical information theory, we know that entropy is a useful measure of the complexity of a random variable and provides a lower bound on the minimum expected number of bits needed for transmitting its state. In this paper, we propose to measure the complexity of a learning task by the minimum expected number of questions that need to be answered to solve the task. For example, the minimum expected number of patches that need to be observed to classify FashionMNIST images. We prove several properties of the proposed complexity measure, including connections with classical entropy and sub-additivity for multiple tasks. As the computation of the minimum expected number of questions is generally intractable, we propose a greedy procedure called “information pursuit” (IP), which selects one question at a time depending on previous questions and their answers. This requires learning a probabilistic generative model relating data and questions to the task, for which we employ variational autoencoders and normalizing flows. We illustrate the usefulness of the proposed measure on various binary image classification tasks using image patches as the query set. Our results indicate that the complexity of a classification task increases as signal-to-noise ratio decreases, and that classification of the KMNIST dataset is more complex than classification of the FashionMNIST dataset. As a byproduct of choosing patches as queries, our approach also provides a principled way of determining which pixels in an image are most informative for a task. | Reject |
ICLR.cc/2019/Conference | Rotation Equivariant Networks via Conic Convolution and the DFT | Performance of neural networks can be significantly improved by encoding known invariance for particular tasks. Many image classification tasks, such as those related to cellular imaging, exhibit invariance to rotation. In particular, to aid convolutional neural networks in learning rotation invariance, we consider a simple, efficient conic convolutional scheme that encodes rotational equivariance, along with a method for integrating the magnitude response of the 2D-discrete-Fourier transform (2D-DFT) to encode global rotational invariance. We call our new method the Conic Convolution and DFT Network (CFNet). We evaluated the efficacy of CFNet as compared to a standard CNN and group-equivariant CNN (G-CNN) for several different image classification tasks and demonstrated improved performance, including classification accuracy, computational efficiency, and its robustness to hyperparameter selection. Taken together, we believe CFNet represents a new scheme that has the potential to improve many imaging analysis applications. | Withdrawn |
ICLR.cc/2019/Conference | SHE2: Stochastic Hamiltonian Exploration and Exploitation for Derivative-Free Optimization | Derivative-free optimization (DFO) using trust region methods is frequently used for machine learning applications, such as (hyper-)parameter optimization without the derivatives of objective functions known. Inspired by the recent work in continuous-time minimizers, our work models the common trust region methods with the exploration-exploitation using a dynamical system coupling a pair of dynamical processes. While the first exploration process searches the minimum of the blackbox function through minimizing a time-evolving surrogation function, another exploitation process updates the surrogation function time-to-time using the points traversed by the exploration process. The efficiency of derivative-free optimization thus depends on ways the two processes couple. In this paper, we propose a novel dynamical system, namely \ThePrev---\underline{S}tochastic \underline{H}amiltonian \underline{E}xploration and \underline{E}xploitation, that surrogates the subregions of blackbox function using a time-evolving quadratic function, then explores and tracks the minimum of the quadratic functions using a fast-converging Hamiltonian system. The \ThePrev\ algorithm is later provided as a discrete-time numerical approximation to the system. To further accelerate optimization, we present \TheName\ that parallelizes multiple \ThePrev\ threads for concurrent exploration and exploitation. Experiment results based on a wide range of machine learning applications show that \TheName\ outperform a boarder range of derivative-free optimization algorithms with faster convergence speed under the same settings. | Reject |
ICLR.cc/2023/Conference | BiasPAD: A Bias-Progressive Auto-Debiasing Framework | While large pre-trained language models have made great strides in natural language understanding benchmarks, recent studies have found that models rely more on the superficial or short-cut features to make predictions. In this paper, we study how to progressively and automatically detect and filter the biased data to train a robust debiased model for NLU tasks. Rather than focusing on the human-predefined biases or biases captured by a bias-only model of limited-capacity, we introduce a new debiasing framework, called Bias-Progressive Auto-Debiasing (BiasPAD), based on two observations: i) the higher the proportion of bias in the training data, the more biased the model will be, and ii) a more biased model has higher confidence in predicting the bias. The framework progressively trains a bias-only model by using the most biased samples detected in the previous epoch, which ensures a more biased model and leads to a robust debiased model. The extensive experiments demonstrate the effectiveness of the proposed framework on several challenging NLU datasets, where on HANS, we achieve 5% accuracy improvement. | Withdrawn |
ICLR.cc/2022/Conference | Dynamic Parameterized Network for CTR Prediction | Learning to capture feature relations effectively and efficiently is essential in click-through rate (CTR) prediction of modern recommendation systems. Most existing CTR prediction methods model such relations either through tedious manually-designed low-order interactions or through inflexible and inefficient high-order interactions, which both require extra DNN modules for implicit interaction modeling. In this paper, we proposed a novel plug-in operation, Dynamic Parameterized Operation (DPO), to learn both explicit and implicit interaction instance-wisely. We showed that the introduction of DPO into DNN modules and Attention modules can respectively benefit two main tasks in CTR prediction, enhancing the adaptiveness of feature-based modeling and improving user behavior modeling with the instance-wise locality. Our Dynamic Parameterized Networks significantly outperforms state-of-the-art methods in the offline experiments on the public dataset and real-world production dataset, together with an online A/B test. Furthermore, the proposed Dynamic Parameterized Networks has been deployed in the ranking system of one of the world's largest e-commerce companies, serving the main traffic of hundreds of millions of active users. | Reject |
ICLR.cc/2021/Conference | Graph Convolutional Value Decomposition in Multi-Agent Reinforcement Learning | We propose a novel framework for value function factorization in multi-agent deep reinforcement learning using graph neural networks (GNNs). In particular, we consider the team of agents as the set of nodes of a complete directed graph, whose edge weights are governed by an attention mechanism. Building upon this underlying graph, we introduce a mixing GNN module, which is responsible for two tasks: i) factorizing the team state-action value function into individual per-agent observation-action value functions, and ii) explicit credit assignment to each agent in terms of fractions of the global team reward. Our approach, which we call GraphMIX, follows the centralized training and decentralized execution paradigm, enabling the agents to make their decisions independently once training is completed. Experimental results on the StarCraft II multi-agent challenge (SMAC) environment demonstrate the superiority of our proposed approach as compared to the state-of-the-art. | Withdrawn |
ICLR.cc/2023/Conference | Contrastive Learning of Molecular Representation with Fragmented Views | Molecular representation learning is a fundamental task for AI-based drug design and discovery. Contrastive learning is an attractive framework for this task, as also evidenced in various domains of representation learning, e.g., image, language, and speech. However, molecule-specific ways of constructing good positive or negative views in contrastive training under consideration of their chemical semantics have been relatively under-explored. In this paper, we consider a molecule as a bag of meaningful fragments, e.g., functional groups, by disconnecting a non-ring single bond as the semantic-preserving transformation. Then, we suggest to construct a complete (or incomplete) bag of fragments as the positive (or negative) views of a molecule: each fragment loses chemical substructures from the original molecule, while the union of the fragments does not. Namely, this provides easy positive and hard negative views simultaneously for contrastive representation learning so that it can selectively learn useful features and ignore nuisance features. Furthermore, we additionally suggest to optimize the torsional angle reconstruction loss around the fragmented bond to incorporate with 3D geometric structure in the pre-training dataset. Our experiments demonstrate that our scheme outperforms prior state-of-the-art molecular representation learning methods across various downstream molecule property prediction tasks. | Reject |
ICLR.cc/2020/Conference | Ellipsoidal Trust Region Methods for Neural Network Training | We investigate the use of ellipsoidal trust region constraints for second-order optimization of neural networks. This approach can be seen as a higher-order counterpart of adaptive gradient methods, which we here show to be interpretable as first-order trust region methods with ellipsoidal constraints. In particular, we show that the preconditioning matrix used in RMSProp and Adam satisfies the necessary conditions for provable convergence of second-order trust region methods with standard worst-case complexities. Furthermore, we run experiments across different neural architectures and datasets to find that the ellipsoidal constraints constantly outperform their spherical counterpart both in terms of number of backpropagations and asymptotic loss value. Finally, we find comparable performance to state-of-the-art first-order methods in terms of backpropagations, but further advances in hardware are needed to render Newton methods competitive in terms of time. | Reject |
ICLR.cc/2021/Conference | Asymptotic Optimality of Self-Representative Low-Rank Approximation and Its Applications | We propose a novel technique for sampling representatives from a large, unsupervised dataset. The approach is based on the concept of {\em self-rank}, defined as the minimum number of samples needed to reconstruct all samples with an accuracy proportional to the rank-$K$ approximation. As the exact computation of self-rank requires a computationally expensive combinatorial search, we propose an efficient algorithm that jointly estimates self-rank and selects the optimal samples with high accuracy. A theoretical upper bound is derived that reaches the tightest bound for two asymptotic cases. The best approximation ratio for self-representative low-rank approximation was presented in ICML 2017~\cite{Chierichetti-icml-2017}, which was further improved by the bound $\sqrt{1+K}$ reported in~NeurIPS 2019~\cite{dan2019optimal}. Both of these bounds depend solely on the number of selected samples. In this paper, for the first time, we present an adaptive approximation ratio depending on spectral properties of the original dataset, $\small{\boldsymbol{A}\in \mathbb{R}^{N\times M}}$. In particular, our performance bound is proportional to the condition number $\kappa(\boldsymbol{A})$. Our derived approximation ratio is expressed as $1+(\kappa(\boldsymbol{A})^2-1)/(N-K)$ which approaches $1$ in two asymptotic cases.
In addition to evaluating the proposed algorithm on a synthetic dataset, we show that the proposed sampling scheme can be utilized in real-world applications such as graph node sampling for optimizing the shortest path criterion, and learning a classifier with sampled data. | Withdrawn |
ICLR.cc/2019/Conference | Self-Tuning Networks: Bilevel Optimization of Hyperparameters using Structured Best-Response Functions | Hyperparameter optimization can be formulated as a bilevel optimization problem, where the optimal parameters on the training set depend on the hyperparameters. We aim to adapt regularization hyperparameters for neural networks by fitting compact approximations to the best-response function, which maps hyperparameters to optimal weights and biases. We show how to construct scalable best-response approximations for neural networks by modeling the best-response as a single network whose hidden units are gated conditionally on the regularizer. We justify this approximation by showing the exact best-response for a shallow linear network with L2-regularized Jacobian can be represented by a similar gating mechanism. We fit this model using a gradient-based hyperparameter optimization algorithm which alternates between approximating the best-response around the current hyperparameters and optimizing the hyperparameters using the approximate best-response function. Unlike other gradient-based approaches, we do not require differentiating the training loss with respect to the hyperparameters, allowing us to tune discrete hyperparameters, data augmentation hyperparameters, and dropout probabilities. Because the hyperparameters are adapted online, our approach discovers hyperparameter schedules that can outperform fixed hyperparameter values. Empirically, our approach outperforms competing hyperparameter optimization methods on large-scale deep learning problems. We call our networks, which update their own hyperparameters online during training, Self-Tuning Networks (STNs). | Accept (Poster) |
ICLR.cc/2022/Conference | Hermitry Ratio: Evaluating the validity of perturbation methods for explainable deep learning | Perturbation methods are model-agnostic methods used to generate heatmaps to explain black-box algorithms such as deep neural networks. Perturbation methods work by perturbing the input image. However, by perturbing parts of the input image we are changing the underlying structure of the image, potentially generating out-of-distribution (OOD) data. This would violate one of the core assumptions in supervised learning, namely that the train and test data come from the same distribution.
In this study, we coin the term hermitry ratio to quantify the utility of perturbation methods by looking at the amount of OOD samples they produce. Using this metric, we observe the utility of XAI methods (Occlusion analysis, LIME, Anchor LIME, Kernel SHAP) for image classification models ResNet50, DensNet121 and MnasNet1.0 on three classes of the ImageNet dataset. Our results show that, to some extent, \emph{all} four perturbation methods generate OOD data regardless of architecture or image class. Occlusion analysis primarily produces in-distribution perturbations while LIME produces mostly OOD perturbations. | Withdrawn |
ICLR.cc/2020/Conference | Actor-Critic Approach for Temporal Predictive Clustering | Due to the wider availability of modern electronic health records (EHR), patient care data is often being stored in the form of time-series. Clustering such time-series data is crucial for patient phenotyping, anticipating patients’ prognoses by identifying “similar” patients, and designing treatment guidelines that are tailored to homogeneous patient subgroups. In this paper, we develop a deep learning approach for clustering time-series data, where each cluster comprises patients who share similar future outcomes of interest (e.g., adverse events, the onset of comorbidities, etc.). The clustering is carried out by using our novel loss functions that encourage each cluster to have homogeneous future outcomes. We adopt actor-critic models to allow “back-propagation” through the sampling process that is required for assigning clusters to time-series inputs. Experiments on two real-world datasets show that our model achieves superior clustering performance over state-of-the-art benchmarks and identifies meaningful clusters that can be translated into actionable information for clinical decision-making. | Reject |
ICLR.cc/2021/Conference | Align-RUDDER: Learning From Few Demonstrations by Reward Redistribution | Reinforcement Learning algorithms require a large number of samples to solve complex tasks with sparse and delayed rewards.
Complex tasks are often hierarchically composed of sub-tasks.
A step in the Q-function indicates solving a sub-task, where the expectation of the return increases.
RUDDER identifies these steps and then redistributes reward to them, thus immediately giving reward if sub-tasks are solved.
Since the delay of rewards is reduced, learning is considerably sped up.
However, for complex tasks, current exploration strategies struggle with discovering episodes with high rewards.
Therefore, we assume that episodes with high rewards are given as demonstrations and do not have to be discovered by exploration.
Typically the number of demonstrations is small and RUDDER's LSTM model does not learn well.
Hence, we introduce Align-RUDDER, which is RUDDER with two major modifications.
First, Align-RUDDER assumes that episodes with high rewards are given as demonstrations,
replacing RUDDER’s safe exploration and lessons replay buffer.
Second, we substitute RUDDER’s LSTM model by a profile model that is obtained from multiple sequence alignment of demonstrations.
Profile models can be constructed from as few as two demonstrations.
Align-RUDDER inherits the concept of reward redistribution, which speeds up learning by reducing the delay of rewards.
Align-RUDDER outperforms competitors on complex artificial tasks with delayed reward and few demonstrations.
On the MineCraft ObtainDiamond task, Align-RUDDER is able to mine a diamond, though not frequently. | Reject |
ICLR.cc/2023/Conference | When do Convolutional Neural Networks Stop Learning? | Convolutional Neural Networks (CNNs) have demonstrated outstanding performance in computer vision tasks such as image classification, detection, segmentation, and medical image analysis. In general, an arbitrary number of epochs is used to train such neural networks. In a single epoch, the entire training data---divided by batch size---are fed to the network. In practice, validation error with training loss is used to estimate the neural network's generalization, which indicates the optimal learning capacity of the network. Current practice is to stop training when the training loss decreases and the gap between training and validation error increases (i.e., the generalization gap) to avoid overfitting. However, this is a trial-and-error-based approach which raises a critical question: Is it possible to estimate when neural networks stop learning based on training data? This research work introduces a hypothesis that analyzes the data variation across all the layers of a CNN variant to anticipate its near-optimal learning capacity. In the training phase, we use our hypothesis to anticipate the near-optimal learning capacity of a CNN variant without using any validation data. Our hypothesis can be deployed as a plug-and-play to any existing CNN variant without introducing additional trainable parameters to the network. We test our hypothesis\footnote{https://github.com/PaperUnderReviewDeepLearning/ \\Optimization} on six different CNN variants and three different datasets (CIFAR10, CIFAR100, and SVHN). The result based on these CNN variants and datasets shows that our hypothesis saves 58.49\% of computational time (on average) in training. | Withdrawn |
ICLR.cc/2022/Conference | Domain-wise Adversarial Training for Out-of-Distribution Generalization | Despite the impressive success on many tasks, deep learning models are shown to rely on spurious features, which will catastrophically fail when generalized to out-of-distribution (OOD) data. To alleviate this issue, Invariant Risk Minimization (IRM) is proposed to extract domain-invariant features for OOD generalization. Nevertheless, recent work shows that IRM is only effective for a certain type of distribution shift (e.g., correlation shift) while fails for other cases (e.g., diversity shift). Meanwhile, another thread of method, Adversarial Training (AT), has shown better domain transfer performance, suggesting that it is potential to be an effective candidate for extracting domain-invariant features. In this paper, we investigate this possibility by exploring the similarity between the IRM and AT objectives. Inspired by this connection, we propose Domain-wise Adversarial Training (DAT), an AT-inspired method for alleviating distribution shift by domain-specific perturbations. Extensive experiments show that our proposed DAT can effectively remove the domain-varying features and improve OOD generalization on both correlation shift and diversity shift tasks. | Reject |
ICLR.cc/2023/Conference | Extending graph transformers with quantum computed aggregation | Recently, efforts have been made in the community to design new Graph Neural Networks (GNN), as limitations of Message Passing Neural Networks became more apparent. This led to the appearance of Graph Transformers using global graph features such as Laplacian Eigenmaps. In our paper, we introduce a GNN architecture where the aggregation weights are computed using the long-range correlations of a quantum system. These correlations are generated by translating the graph topology into the interactions of a set of qubits in a quantum computer. The recent development of quantum processing units enables the computation of a new family of global graph features that would be otherwise out of reach for classical hardware. We give some theoretical insights about the potential benefits of this approach, and benchmark our algorithm on standard datasets. Although not being adapted to all datasets, our model performs similarly to standard GNN architectures, and paves a promising future for quantum enhanced GNNs. | Reject |
ICLR.cc/2023/Conference | Multi-Modality Alone is Not Enough: Generating Scene Graphs using Cross-Relation-Modality Tokens | Recent years have seen a growing interest in Scene Graph Generation (SGG), a comprehensive visual scene understanding task that aims to predict the relationships between objects detected in a scene. One of its key challenges is the strong bias of the visual world around us toward a few frequently occurring relationships, leaving a long tail of under-represented classes. Although infusing additional modalities is one prominent way to improve SGG performance on under-represented classes, we argue that using additional modalities alone is not enough. We propose to inject entity relation information (Cross-Relation) and modality dependencies (Cross-Modality) into each embedding token of a transformer which we term primal fusion. The resulting Cross-RElAtion-Modality (CREAM) token acts as a strong inductive bias for the SGG framework. Our experimental results on the Visual Genome dataset demonstrate that our CREAM model outperforms state-of-the-art SGG models by around 20% while being simpler and requiring substantially less computation. Additionally, to analyse the generalisability of the CREAM model we also evaluate it on the Open Images dataset. Finally, we examine the impact of the depth-map quality on SGG performance and empirically show the superiority of our model over the prior state of the art by better capturing the depth data, boosting the performance by a margin of around 25%. | Reject |
ICLR.cc/2023/Conference | Quasi-optimal Reinforcement Learning with Continuous Actions | Many real-world applications of reinforcement learning (RL) require making decisions in continuous action environments. In particular, determining the optimal dose level plays a vital role in developing medical treatment regimes. One challenge in adapting existing RL algorithms to medical applications, however, is that the popular infinite support stochastic policies, e.g., Gaussian policy, may assign riskily high dosages and harm patients seriously. Hence, it is important to induce a policy class whose support only contains near-optimal actions, and shrink the action-searching area for effectiveness and reliability. To achieve this, we develop a novel quasi-optimal learning algorithm, which can be easily optimized in off-policy settings with guaranteed convergence under general function approximations. Theoretically, we analyze the consistency, sample complexity, adaptability, and convergence of the proposed algorithm. We evaluate our algorithm with comprehensive simulated experiments and a dose suggestion real application to Ohio Type 1 diabetes dataset. | Accept: poster |
ICLR.cc/2022/Conference | An Investigation on Hardware-Aware Vision Transformer Scaling | Vision Transformer (ViT) has demonstrated promising performance in various computer vision tasks, and recently attracted a lot of research attention. Many recent works have focused on proposing new architectures to improve ViT and deploying it into real-world applications. However, little effort has been made to analyze and understand ViT’s architecture design space and its implication of hardware-cost on different devices. In this work, by simply scaling ViT's depth, width, input size, and other basic configurations, we show that a scaled vanilla ViT model without bells and whistles can achieve comparable or superior accuracy-efficiency trade-off than most of the latest ViT variants. Specifically, compared to DeiT-Tiny, our scaled model achieves a $\uparrow1.9\%$ higher ImageNet top-1 accuracy under the same FLOPs and a $\uparrow3.7\%$ better ImageNet top-1 accuracy under the same latency on an NVIDIA Edge GPU TX2. Motivated by this, we further investigate the extracted scaling strategies from the following two aspects: (1) "can these scaling strategies be transferred across different real hardware devices?''; and (2) "can these scaling strategies be transferred to different ViT variants and tasks?''. For (1), our exploration, based on various devices with different resource budgets, indicates that the transferability effectiveness depends on the underlying device together with its corresponding deployment tool; for (2), we validate the effective transferability of the aforementioned scaling strategies obtained from a vanilla ViT model on top of an image classification task to the PiT model, a strong ViT variant targeting efficiency, as well as object detection and video classification tasks. In particular, when transferred to PiT, our scaling strategies lead to a boosted ImageNet top-1 accuracy of from $74.6\%$ to $76.7\%$ ($\uparrow2.1\%$) under the same 0.7G FLOPs; and when transferred to the COCO object detection task, the average precision is boosted by $\uparrow0.7\%$ under a similar throughput on a V100 GPU. | Reject |
ICLR.cc/2020/Conference | Scaling Autoregressive Video Models | Due to the statistical complexity of video, the high degree of inherent stochasticity, and the sheer amount of data, generating natural video remains a challenging task. State-of-the-art video generation models attempt to address these issues by combining sometimes complex, often video-specific neural network architectures, latent variable models, adversarial training and a range of other methods. Despite their often high complexity, these approaches still fall short of generating high quality video continuations outside of narrow domains and often struggle with fidelity. In contrast, we show that conceptually simple, autoregressive video generation models based on a three-dimensional self-attention mechanism achieve highly competitive results across multiple metrics on popular benchmark datasets for which they produce continuations of high fidelity and realism. Furthermore, we find that our models are capable of producing diverse and surprisingly realistic continuations on a subset of videos from Kinetics, a large scale action recognition dataset comprised of YouTube videos exhibiting phenomena such as camera movement, complex object interactions and diverse human movement. To our knowledge, this is the first promising application of video-generation models to videos of this complexity. | Accept (Spotlight) |
ICLR.cc/2020/Conference | Adversarial Filters of Dataset Biases | Large-scale benchmark datasets have been among the major driving forces in AI, supporting training of models and measuring their progress. The key assumption is that these benchmarks are realistic approximations of the target tasks in the real world. However, while machine performance on these benchmarks advances rapidly --- often surpassing human performance --- it still struggles on the target tasks in the wild. This raises an important question: whether the surreal high performance on existing benchmarks are inflated due to spurious biases in them, and if so, how we can effectively revise these benchmarks to better simulate more realistic problem distributions in the real world.
In this paper, we posit that while the real world problems consist of a great deal of long-tail problems, existing benchmarks are overly populated with a great deal of similar (thus non-tail) problems, which in turn, leads to a major overestimation of true AI performance. To address this challenge, we present a novel framework of Adversarial Filters to investigate model-based reduction of dataset biases. We discuss that the optimum bias reduction via AFOptimum is intractable, thus propose AFLite, an iterative greedy algorithm that adversarially filters out data points to identify a reduced dataset with more realistic problem distributions and considerably less spurious biases.
AFLite is lightweight and can in principle be applied to any task and dataset. We apply it to popular benchmarks that are practically solved --- ImageNet and Natural Language Inference (SNLI, MNLI, QNLI) --- and present filtered counterparts as new challenge datasets where the model performance drops considerably (e.g., from 84% to 24% for ImageNet and from 92% to 62% for SNLI), while human performance remains high. An extensive suite of analysis demonstrates that AFLite effectively reduces measurable dataset biases in both the synthetic and real datasets. Finally, we introduce new measures of dataset biases based on K-nearest-neighbors to help guide future research on dataset developments and bias reduction. | Reject |
ICLR.cc/2022/Conference | Spending Your Winning Lottery Better After Drawing It | Lottery Ticket Hypothesis (LTH) (Frankle & Carbin, 2019) suggest suggests that a dense neural network contains a sparse sub-network that can match the performance of the original dense network when trained in isolation from scratch. Most works retrain the sparse sub-network with the same training protocols as its dense network, such as initialization, architecture blocks, and training recipes. However, till now it is unclear that whether these training protocols are optimal for sparse networks.
In this paper, we demonstrate that it is unnecessary for spare retraining to strictly inherit those properties from the dense network. Instead, by plugging in purposeful "tweaks" of the sparse subnetwork architecture or its training recipe, its retraining can be significantly improved than the default, especially at high sparsity levels. Combining all our proposed "tweaks" can yield the new state-of-the-art performance of LTH, and these modifications can be easily adapted to other sparse training algorithms in general. Specifically, we have achieved a significant and consistent performance gain of 1.05% - 4.93% for ResNet18 on CIFAR-100 over vanilla-LTH. Moreover, our methods are shown to generalize across datasets (CIFAR10, CIFAR100, TinyImageNet) and architectures (Vgg16, ResNet-18/ResNet-34, MobileNet). All codes will be publicly available. | Withdrawn |
ICLR.cc/2021/Conference | Domain-slot Relationship Modeling using a Pre-trained Language Encoder for Multi-Domain Dialogue State Tracking | Dialogue state tracking for multi-domain dialogues is challenging because the model should be able to track dialogue states across multiple domains and slots. Past studies had its limitations in that they did not factor in the relationship among different domain-slot pairs. Although recent approaches did support relationship modeling among the domain-slot pairs, they did not leverage a pre-trained language model, which has improved the performance of numerous natural language tasks, in the encoding process. Our approach fills the gap between these previous studies. We propose a model for multi-domain dialogue state tracking that effectively models the relationship among domain-slot pairs using a pre-trained language encoder. Inspired by the way the special $[CLS]$ token in BERT is used to aggregate the information of the whole sequence, we use multiple special tokens for each domain-slot pair that encodes information corresponding to its domain and slot. The special tokens are run together with the dialogue context through the pre-trained language encoder, which effectively models the relationship among different domain-slot pairs. Our experimental results show that our model achieves state-of-the-art performance on the MultiWOZ-2.1 and MultiWOZ-2.2 dataset. | Reject |
ICLR.cc/2021/Conference | Learning to Observe with Reinforcement Learning | We consider a decision making problem where an autonomous agent decides on which actions to take based on the observations it collects from the environment. We are interested in revealing the information structure of the observation space illustrating which type of observations are the most important (such as position versus velocity) and the dependence of this on the state of agent (such as at the bottom versus top of a hill). We approach this problem by associating a cost with collecting observations which increases with the accuracy. We adopt a reinforcement learning (RL) framework where the RL agent learns to adjust the accuracy of the observations alongside learning to perform the original task. We consider both the scenario where the accuracy can be adjusted continuously and also the scenario where the agent has to choose between given preset levels, such as taking a sample perfectly or not taking a sample at all. In contrast to the existing work that mostly focuses on sample efficiency during training, our focus is on the behaviour during the actual task. Our results illustrate that the RL agent can learn to use the observation space efficiently and obtain satisfactory performance in the original task while collecting effectively smaller amount of data. By uncovering the relative usefulness of different types of observations and trade-offs within, these results also provide insights for further design of active data acquisition schemes. | Reject |
ICLR.cc/2020/Conference | SoftAdam: Unifying SGD and Adam for better stochastic gradient descent | Abstract Stochastic gradient descent (SGD) and Adam are commonly used to optimize deep neural networks, but choosing one usually means making tradeoffs between speed, accuracy and stability. Here we present an intuition for why the tradeoffs exist as well as a method for unifying the two in a continuous way. This makes it possible to control the way models are trained in much greater detail. We show that for default parameters, the new algorithm equals or outperforms SGD and Adam across a range of models for image classification tasks and outperforms SGD for language modeling tasks. | Reject |
ICLR.cc/2022/Conference | CycleMLP: A MLP-like Architecture for Dense Prediction | This paper presents a simple MLP-like architecture, CycleMLP, which is a versatile backbone for visual recognition and dense predictions. As compared to modern MLP architectures, e.g. , MLP-Mixer, ResMLP, and gMLP, whose architectures are correlated to image size and thus are infeasible in object detection and segmentation, CycleMLP has two advantages compared to modern approaches. (1) It can cope
with various image sizes. (2) It achieves linear computational complexity to image size by using local windows. In contrast, previous MLPs have $O(N^2)$ computations due to fully spatial connections. We build a family of models which surpass existing MLPs and even state-of-the-art Transformer-based models, e.g. Swin Transformer, while using fewer parameters and FLOPs. We expand the MLP-like models’ applicability, making them a versatile backbone for dense prediction tasks. CycleMLP achieves competitive results on object detection, instance segmentation, and semantic segmentation. In particular, CycleMLP-Tiny outperforms Swin-Tiny by 1.3% mIoU on ADE20K dataset with fewer FLOPs. Moreover, CycleMLP also shows excellent zero-shot robustness on ImageNet-C dataset. | Accept (Oral) |
ICLR.cc/2019/Conference | COLLABORATIVE MULTIAGENT REINFORCEMENT LEARNING IN HOMOGENEOUS SWARMS | A deep reinforcement learning solution is developed for a collaborative multiagent system. Individual agents choose actions in response to the state of the environment, their own state, and possibly partial information about the state of other agents. Actions are chosen to maximize a collaborative long term discounted reward that encompasses the individual rewards collected by each agent. The paper focuses on developing a scalable approach that applies to large swarms of homogeneous agents. This is accomplished by forcing the policies of all agents to be the same resulting in a constrained formulation in which the experiences of each agent inform the learning process of the whole team, thereby enhancing the sample efficiency of the learning process. A projected coordinate policy gradient descent algorithm is derived to solve the constrained reinforcement learning problem. Experimental evaluations in collaborative navigation, a multi-predator-multi-prey game, and a multiagent survival game show marked improvements relative to methods that do not exploit the policy equivalence that naturally arises in homogeneous swarms. | Reject |
ICLR.cc/2023/Conference | Cooperate or Compete: A New Perspective on Training of Generative Networks | GANs have two competing modules: the generator module is trained to generate new examples, and the discriminator module is trained to discriminate real examples from generated examples. The training procedure of GAN is modeled as a finitely repeated simultaneous game. Each module tries to increase its performance at every repetition of the base game (at every batch of training data) in a non-cooperative manner. We observed that each module can perform better and learn faster if training is modeled as an infinitely repeated simultaneous game. At every repetition of the base game (at every batch of training data) the stronger module (whose performance is increased or remains the same compared to the previous batch of training data) cooperates with the weaker module (whose performance is decreased compared to the previous batch of training data) and only the weaker module is allowed to increase its performance. | Desk_Rejected |
ICLR.cc/2022/Conference | Modular Lifelong Reinforcement Learning via Neural Composition | Humans commonly solve complex problems by decomposing them into easier subproblems and then combining the subproblem solutions. This type of compositional reasoning permits reuse of the subproblem solutions when tackling future tasks that share part of the underlying compositional structure. In a continual or lifelong reinforcement learning (RL) setting, this ability to decompose knowledge into reusable components would enable agents to quickly learn new RL tasks by leveraging accumulated compositional structures. We explore a particular form of composition based on neural modules and present a set of RL problems that intuitively admit compositional solutions. Empirically, we demonstrate that neural composition indeed captures the underlying structure of this space of problems. We further propose a compositional lifelong RL method that leverages accumulated neural components to accelerate the learning of future tasks while retaining performance on previous tasks via off-line RL over replayed experiences. | Accept (Poster) |
ICLR.cc/2020/Conference | Three-Head Neural Network Architecture for AlphaZero Learning | The search-based reinforcement learning algorithm AlphaZero has been used as a general method for
mastering two-player games Go, chess and Shogi. One crucial ingredient in AlphaZero (and its predecessor AlphaGo Zero) is the two-head network architecture that outputs two estimates --- policy and value --- for one input game state. The merit of such an architecture is that letting policy and value learning share the same representation substantially improved generalization of the neural net.
A three-head network architecture has been recently proposed that can learn a third action-value head on a fixed dataset the same as for two-head net. Also, using the action-value head in Monte Carlo tree search (MCTS) improved the search efficiency.
However, effectiveness of the three-head network has not been investigated in an AlphaZero style learning paradigm.
In this paper, using the game of Hex as a test domain, we conduct an empirical study of the three-head network architecture in AlpahZero learning. We show that the architecture is also advantageous at the zero-style iterative learning. Specifically, we find that three-head network can induce the following benefits: (1) learning can become faster as search takes advantage of the additional action-value head; (2) better prediction results than two-head architecture can be achieved when using additional action-value learning as an auxiliary task. | Reject |
ICLR.cc/2023/Conference | PA-LoFTR: Local Feature Matching with 3D Position-Aware Transformer | We propose a novel image feature matching method that utilizes 3D position information to augment feature representation with a deep neural network. The proposed method introduces 3D position embedding to a state-of-the-art feature matcher, LoFTR, and achieves more promising performance. Following the coarse-to-fine matching pipeline of LoFTR, we construct a Transformer-based neural network that generates dense pixel-wise matches. Instead of using 2D position embeddings for transformer, the proposed method generates 3D position embeddings that can precisely capture position correspondence of matches between images. Importantly, in order to guide neural network to learn 3D space information, we augment features with depth information generated by a depth predictor. In this way, our method, PA-LoFTR, can generate 3D position-aware local feature descriptors with Transformer. We experiment on indoor datasets, and results show that PA-LoFTR improves the performance of feature matching compared to state-of-the-art methods. | Reject |
ICLR.cc/2020/Conference | Improving Visual Relation Detection using Depth Maps | State of the art visual relation detection methods mostly rely on object information extracted from RGB images such as predicted class probabilities, 2D bounding boxes and feature maps. In this paper, we argue that the 3D positions of objects in space can provide additional valuable information about object relations. This information helps not only to detect spatial relations, such as \textit{standing behind}, but also non-spatial relations, such as \textit{holding}. Since 3D information of a scene is not easily accessible, we propose incorporating a pre-trained RGB-to-Depth model within visual relation detection frameworks. We discuss different feature extraction strategies from depth maps and show their critical role in relation detection.
Our experiments confirm that the performance of state-of-the-art visual relation detection approaches can significantly be improved by utilizing depth map information. | Reject |
ICLR.cc/2023/Conference | Repository-Level Prompt Generation for Large Language Models of Code | With the success of large language models (LLMs) of code and their use as code assistants (e.g.\ Codex used in GitHub Copilot, techniques for introducing domain-specific knowledge in the prompt design process become important. In this work, we propose a framework called Repo-Level Prompt Generator that learns to generate example-specific prompts using prompt proposals. The prompt proposals take context from the entire repository, thereby incorporating both the structure of the repository and the context from other relevant files (e.g.\ imports, parent class files). Our technique doesn't require any access to the weights of the LLM, making it applicable in cases where we only have black-box access to the LLM. We conduct experiments on the task of single-line code-autocompletion using code repositories taken from Google Code archives. We demonstrate that an oracle constructed from our prompt proposals gives a remarkably high relative improvement of 36\% over Codex, showing the quality of these proposals. Further, we show that when we train a model to select the best prompt proposal, we can achieve significant performance gains over Codex and other baselines. | Reject |
ICLR.cc/2023/Conference | Exploring Neural Network Representational Similarity using Filter Subspaces | Analyzing representational similarity in neural networks is crucial to numerous tasks, such as interpreting or transferring deep models. One typical approach is to input probing data into convolutional neural networks (CNNs) as stimuli to reveal their deep representation for model similarity analysis. Those methods are often computationally expensive and stimulus-dependent. By representing filter subspace in a CNN as a set of filter atoms, previous work has reported competitive performance in continual learning by learning a different set of filter atoms for each task while sharing common atom coefficients across tasks. Inspired by this observation, in this paper, we propose a new paradigm for reducing representational similarity analysis in CNNs to filter subspace distance assessment. Specifically, when filter atom coefficients are shared across networks, model representational similarity can be significantly simplified as calculating the cosine distance among respective filter atoms, to achieve \textit{millions of times} computation reduction. We provide both theoretical and empirical evidence that this simplified filter subspace-based similarity preserves a strong linear correlation with other popular stimulus-based metrics, while being significantly more efficient and robust to probing data. We further validate the effectiveness of the proposed method in various applications, such as analyzing training dynamics as well as in federated and continual learning. We hope our findings can help further explorations of real-time large-scale representational similarity analysis in neural networks. | Reject |
ICLR.cc/2020/Conference | Learning an off-policy predictive state representation for deep reinforcement learning for vision-based steering in autonomous driving | An algorithm is introduced for learning a predictive state representation with off-policy temporal difference (TD) learning that is then used to learn to steer a vehicle with reinforcement learning. There are three components being learned simultaneously: (1) the off-policy predictions as a compact representation of state, (2) the behavior policy distribution for estimating the off-policy predictions, and (3) the deterministic policy gradient for learning to act. A behavior policy discriminator is learned and used for estimating the important sampling ratios needed to learn the predictive representation off-policy with general value functions (GVFs). A linear deterministic policy gradient method is used to train the agent with only the predictive representations while the predictions are being learned. All three components are combined, demonstrated and evaluated on the problem of steering the vehicle from images in the TORCS racing simulator environment.
Steering from only images is a challenging problem where evaluation is completed on a held-out set of tracks that were never seen during training in order to measure the generalization of the predictions and controller. Experiments show the proposed method is able to steer smoothly and navigate many but not all of the tracks available in TORCS with performance that exceeds DDPG using only images as input and approaches the performance of an ideal non-vision based kinematics model. | Withdrawn |
ICLR.cc/2023/Conference | Black-Box Adversarial Attack Guided by Model Behavior for Programming Pre-trained Language Models | Pre-trained models for programming languages are widely used to solve code tasks in Software Engineering (SE) community, such as code clone detection and bug identification. Reliability is the primary concern of these machine learning applications in SE because software failure can lead to intolerable loss. However, deep neural networks are known to suffer from adversarial attacks. In this paper, we propose a novel black-box adversarial attack based on model behaviors for pre-trained programming language models, named Representation Nearest Neighbor Search(RNNS). The proposed approach can efficiently identify adversarial examples via variable replacement in an ample search space of real variable names under similarity constraints. We evaluate RNNS on 6 code tasks (e.g., clone detection), 3 programming languages (Java, Python, and C), and 3 pre-trained code models: CodeBERT, GraphCodeBERT, and CodeT5. The results demonstrate that RNNS outperforms the state-of-the-art black-box attacking method (MHM) in terms of both attack success rate and quality of generated adversarial examples. | Withdrawn |
ICLR.cc/2023/Conference | Reconciling Security and Communication Efficiency in Federated Learning | Cross-device Federated Learning is an increasingly popular machine learning setting to train a model by leveraging a large population of client devices with high privacy and security guarantees. However, communication efficiency remains a major bottleneck when scaling federated learning to production environments, particularly due to bandwidth constraints during uplink communication. In this paper, we formalize and address the problem of compressing client-to-server model updates under the Secure Aggregation primitive, a core component of Federated Learning pipelines that allows the server to aggregate the client updates without accessing them individually. In particular, we adapt standard scalar quantization and pruning methods to Secure Aggregation and propose Secure Indexing, a variant of Secure Aggregation that supports quantization for extreme compression. We establish state-of-the-art results on LEAF benchmarks in a secure Federated Learning setup with up to 40x compression in uplink communication with no meaningful loss in utility compared to uncompressed baselines. | Reject |
ICLR.cc/2019/Conference | Visual Imitation Learning with Recurrent Siamese Networks | People are incredibly skilled at imitating others by simply observing them. They achieve this even in the presence of significant morphological differences and capabilities. Further, people are able to do this from raw perceptions of the actions of others, without direct access to the abstracted demonstration actions and with only partial state information. People therefore solve a difficult problem of understanding the salient features of both observations of others and the relationship to their own state when learning to imitate specific tasks.
However, we can attempt to reproduce a similar demonstration via trail and error and through this gain more understanding of the task space.
To reproduce this ability an agent would need to both learn how to recognize the differences between itself and some demonstration and at the same time learn to minimize the distance between its own performance and that of the demonstration.
In this paper we propose an approach using only visual information to learn a distance metric between agent behaviour and a given video demonstration.
We train an RNN-based siamese model to compute distances in space and time between motion clips while training an RL policy to minimize this distance.
Furthermore, we examine a particularly challenging form of this problem where the agent must learn an imitation based task given a single demonstration.
We demonstrate our approach in the setting of deep learning based control for physical simulation of humanoid walking in both 2D with $10$ degrees of freedom (DoF) and 3D with $38$ DoF. | Reject |
ICLR.cc/2020/Conference | N-BEATS: Neural basis expansion analysis for interpretable time series forecasting | We focus on solving the univariate times series point forecasting problem using deep learning. We propose a deep neural architecture based on backward and forward residual links and a very deep stack of fully-connected layers. The architecture has a number of desirable properties, being interpretable, applicable without modification to a wide array of target domains, and fast to train. We test the proposed architecture on several well-known datasets, including M3, M4 and TOURISM competition datasets containing time series from diverse domains. We demonstrate state-of-the-art performance for two configurations of N-BEATS for all the datasets, improving forecast accuracy by 11% over a statistical benchmark and by 3% over last year's winner of the M4 competition, a domain-adjusted hand-crafted hybrid between neural network and statistical time series models. The first configuration of our model does not employ any time-series-specific components and its performance on heterogeneous datasets strongly suggests that, contrarily to received wisdom, deep learning primitives such as residual blocks are by themselves sufficient to solve a wide range of forecasting problems. Finally, we demonstrate how the proposed architecture can be augmented to provide outputs that are interpretable without considerable loss in accuracy. | Accept (Poster) |
ICLR.cc/2023/Conference | Self-supervised debiasing using low rank regularization | Spurious correlations can cause strong biases in deep neural networks, impairing generalization ability. While most of existing debiasing methods require full supervisions on either spurious attributes or target labels, training a debiased model from a limited amount of both annotations is still an open issue. To overcome such limitations, we first examined an interesting phenomenon by the spectral analysis of latent representations: spuriously correlated, easy-to-learn attributes make neural networks inductively biased towards encoding lower effective rank representations. We also show that a rank regularization can amplify this bias in a way that encourages highly correlated features. Motivated by these observations, we propose a self-supervised debiasing framework that is potentially compatible with unlabeled samples. We first pretrain a biased encoder in a self-supervised manner with the rank regularization, serving as a semantic bottleneck to enforce the encoder to learn the spuriously correlated attributes. This biased encoder is then used to discover and upweight bias-conflicting samples in a downstream task, serving as a boosting to effectively debias the main model. Remarkably, the proposed debiasing framework significantly improves the generalization performance of self-supervised learning baselines and, in some cases, even outperforms state-of-the-art supervised debiasing approaches. | Reject |
ICLR.cc/2021/Conference | Self-supervised Adversarial Robustness for the Low-label, High-data Regime | Recent work discovered that training models to be invariant to adversarial perturbations requires substantially larger datasets than those required for standard classification. Perhaps more surprisingly, these larger datasets can be "mostly" unlabeled. Pseudo-labeling, a technique simultaneously pioneered by four separate and simultaneous works in 2019, has been proposed as a competitive alternative to labeled data for training adversarially robust models. However, when the amount of labeled data decreases, the performance of pseudo-labeling catastrophically drops, thus questioning the theoretical insights put forward by Uesato et al. (2019), which suggest that the sample complexity for learning an adversarially robust model from unlabeled data should match the fully supervised case. We introduce Bootstrap Your Own Robust Latents (BYORL), a self-supervised learning technique based on BYOL for training adversarially robust models. Our method enables us to train robust representations without any labels (reconciling practice with theory). Most notably, this robust representation can be leveraged by a linear classifier to train adversarially robust models, even when the linear classifier is not trained adversarially. We evaluate BYORL and pseudo-labeling on CIFAR-10 and ImageNet and demonstrate that BYORL achieves significantly higher robustness (i.e., models resulting from BYORL are up to two times more accurate). Experiments on CIFAR-10 against $\ell_2$ and $\ell_\infty$ norm-bounded perturbations demonstrate that BYORL achieves near state-of-the-art robustness with as little as 500 labeled examples. We also note that against $\ell_2$ norm-bounded perturbations of size $\epsilon = 128/255$, BYORL surpasses the known state-of-the-art with an accuracy under attack of 77.61% (against 72.91% for the prior art). | Accept (Poster) |
ICLR.cc/2021/Conference | Self-Activating Neural Ensembles for Continual Reinforcement Learning | The ability for an agent to continuously learn new skills without catastrophically forgetting existing knowledge is of critical importance for the development of generally intelligent agents. Most methods devised to address this problem depend heavily on well-defined task boundaries which simplify the problem considerably. Our task-agnostic method, Self-Activating Neural Ensembles (SANE), uses a hierarchical modular architecture designed to avoid catastrophic forgetting without making any such assumptions. At each timestep a path through the SANE tree is activated; during training only activated nodes are updated, ensuring that unused nodes do not undergo catastrophic forgetting. Additionally, new nodes are created as needed, allowing the system to leverage and retain old skills while growing and learning new ones. We demonstrate our approach on MNIST and a set of grid world environments, demonstrating that SANE does not undergo catastrophic forgetting where existing methods do. | Reject |
ICLR.cc/2023/Conference | SAE: Estimation for Transition Matrix in Annotation Algorithms | The transition matrix plays a critical role in label-noise learning tasks, which refers to the transition from clean labels to noisy labels. The majority of recent methods for inferring the transition matrix concentrate on the manually hand-crafted label noise but with bearing the high cost of time and labor. In light of this, several straightforward and effective algorithms are introduced for automatically annotating the label noise. However, the automatic annotation algorithms easily generate wrong pseudo labels for similar semantic categories. Moreover, a special instance-dependent transition matrix is launched due to the mapping from a specific category to other similar categories during the annotation process. To address this issue, we propose a semantic adaption estimator (SAE) to indirectly infer the instance-dependent transition matrix. Specifically, we decouple the original instance-dependent transition matrix to several easy-to-estimate semantic-dependent transition matrices by introducing a semantic adaption loss function. In this way, the original datasets can be decoupled into some simple semantic regions. Then the instance-dependent transition matrix can be built from multiple learned semantic-dependent matrices. Empirical evaluations on two real-world datasets (i.e., S3DIS and ScanNet) demonstrate the superior performance of our method, in comparison with the state-of-the-art. | Withdrawn |
ICLR.cc/2018/Conference | Representing dynamically: An active process for describing sequential data | We propose an unsupervised method for building dynamic representations of sequential data, particularly of observed interactions. The method simultaneously acquires representations of input data and its dynamics. It is based on a hierarchical generative model composed of two levels. In the first level, a model learns representations to generate observed data. In the second level, representational states encode the dynamics of the lower one. The model is designed as a Bayesian network with switching variables represented in the higher level, and which generates transition models. The method actively explores the latent space guided by its knowledge and the uncertainty about it. That is achieved by updating the latent variables from prediction error signals backpropagated to the latent space. So, no encoder or inference models are used since the generators also serve as their inverse transformations.
The method is evaluated in two scenarios, with static images and with videos. The results show that the adaptation over time leads to better performance than with similar architectures without temporal dependencies, e.g., variational autoencoders. With videos, it is shown that the system extracts the dynamics of the data in states that highly correlate with the ground truth of the actions observed. | Reject |
ICLR.cc/2023/Conference | PD-MORL: Preference-Driven Multi-Objective Reinforcement Learning Algorithm | Multi-objective reinforcement learning (MORL) approaches have emerged to tackle many real-world problems with multiple conflicting objectives by maximizing a joint objective function weighted by a preference vector. These approaches find fixed customized policies corresponding to preference vectors specified during training. However, the design constraints and objectives typically change dynamically in real-life scenarios. Furthermore, storing a policy for each potential preference is not scalable. Hence, obtaining a set of Pareto front solutions for the entire preference space in a given domain with a single training is critical. To this end, we propose a novel MORL algorithm that trains a single universal network to cover the entire preference space scalable to continuous robotic tasks. The proposed approach, Preference-Driven MORL (PD-MORL), utilizes the preferences as guidance to update the network parameters. It also employs a novel parallelization approach to increase sample efficiency. We show that PD-MORL achieves up to 25% larger hypervolume for challenging continuous control tasks and uses an order of magnitude fewer trainable parameters compared to prior approaches. | Accept: poster |
ICLR.cc/2022/Conference | Learning Continuous Environment Fields via Implicit Functions | We propose a novel scene representation that encodes reaching distance -- the distance between any position in the scene to a goal along a feasible trajectory. We demonstrate that this environment field representation can directly guide the dynamic behaviors of agents in 2D mazes or 3D indoor scenes. Our environment field is a continuous representation and learned via a neural implicit function using discretely sampled training data. We showcase its application for agent navigation in 2D mazes, and human trajectory prediction in 3D indoor environments. To produce physically plausible and natural trajectories for humans, we additionally learn a generative model that predicts regions where humans commonly appear, and enforce the environment field to be defined within such regions. Extensive experiments demonstrate that the proposed method can generate both feasible and plausible trajectories efficiently and accurately. | Accept (Poster) |
ICLR.cc/2021/Conference | Learning Value Functions in Deep Policy Gradients using Residual Variance | Policy gradient algorithms have proven to be successful in diverse decision making and control tasks. However, these methods suffer from high sample complexity and instability issues. In this paper, we address these challenges by providing a different approach for training the critic in the actor-critic framework. Our work builds on recent studies indicating that traditional actor-critic algorithms do not succeed in fitting the true value function, calling for the need to identify a better objective for the critic. In our method, the critic uses a new state-value (resp. state-action-value) function approximation that learns the value of the states (resp. state-action pairs) relative to their mean value rather than the absolute value as in conventional actor-critic. We prove the theoretical consistency of the new gradient estimator and observe dramatic empirical improvement across a variety of continuous control tasks and algorithms. Furthermore, we validate our method in tasks with sparse rewards, where we provide experimental evidence and theoretical insights. | Accept (Poster) |
ICLR.cc/2021/Conference | Multiplicative Filter Networks | Although deep networks are typically used to approximate functions over high dimensional inputs, recent work has increased interest in neural networks as function approximators for low-dimensional-but-complex functions, such as representing images as a function of pixel coordinates, solving differential equations, or representing signed distance fields or neural radiance fields. Key to these recent successes has been the use of new elements such as sinusoidal nonlinearities, or Fourier features in positional encodings, which vastly outperform simple ReLU networks. In this paper, we propose and empirically demonstrate that an arguably simpler class of function approximators can work just as well for such problems: multiplicative filter networks. In these networks, we avoid traditional compositional depth altogether, and simply multiply together (linear functions of) sinusoidal or Gabor wavelet functions applied to the input. This representation has the notable advantage that the entire function can simply be viewed as a linear function approximator over an exponential number of Fourier or Gabor basis functions, respectively. Despite this simplicity, when compared to recent approaches that use Fourier features with ReLU networks or sinusoidal activation networks, we show that these multiplicative filter networks largely outperform or match the performance of these recent approaches on the domains highlighted in these past works. | Accept (Poster) |
ICLR.cc/2023/Conference | Learning topology-preserving data representations | We propose a method for learning topology-preserving data representations (dimensionality reduction). The method aims to provide topological similarity between the data manifold and its latent representation via enforcing the similarity in topological features (clusters, loops, 2D voids, etc.) and their localization.
The core of the method is the minimization of the Representation Topology Divergence (RTD) between original high-dimensional data and low-dimensional representation in latent space. RTD minimization provides closeness in topological features with strong theoretical guarantees.
We develop a scheme for RTD differentiation and apply it as a loss term for the autoencoder. The proposed method "RTD-AE" better preserves the global structure and topology of the data manifold than state-of-the-art competitors as measured by linear correlation, triplet distance ranking accuracy, and Wasserstein distance between persistence barcodes. | Accept: poster |
ICLR.cc/2022/Conference | Patches Are All You Need? | Although convolutional networks have been the dominant architecture for vision tasks for many years, recent experiments have shown that Transformer-based models, most notably the Vision Transformer (ViT), may exceed their performance in some settings. However, due to the quadratic runtime of the self-attention layers in Transformers, ViTs require the use of patch embeddings, which group together small regions of the image into single input features, in order to be applied to larger image sizes. This raises a question: Is the performance of ViTs due to the inherently-more-powerful Transformer architecture, or is it at least partly due to using patches as the input representation? In this paper, we present some evidence for the latter: specifically, we propose the ConvMixer, an extremely simple model that is similar in spirit to the ViT and the even-more-basic MLP-Mixer in that it operates directly on patches as input, separates the mixing of spatial and channel dimensions, and maintains equal size and resolution throughout the network. In contrast, however, the ConvMixer uses only standard convolutions to achieve the mixing steps. Despite its simplicity, we show that the ConvMixer outperforms the ViT, MLP-Mixer, and some of their variants for similar parameter counts and data set sizes, in addition to outperforming classical vision models such as the ResNet. Our code is available at https://github.com/tmp-iclr/convmixer. | Reject |
ICLR.cc/2020/Conference | CGT: Clustered Graph Transformer for Urban Spatio-temporal Prediction | Deep learning based approaches have been widely used in various urban spatio-temporal forecasting problems, but most of them fail to account for the unsmoothness issue of urban data in their architecture design, which significantly deteriorates their prediction performance. The aim of this paper is to develop a novel clustered graph transformer framework that integrates both graph attention network and transformer under an encoder-decoder architecture to address such unsmoothness issue. Specifically, we propose two novel structural components to refine the architectures of those existing deep learning models. In spatial domain, we propose a gradient-based clustering method to distribute different feature extractors to regions in different contexts. In temporal domain, we propose to use multi-view position encoding to address the periodicity and closeness of urban time series data. Experiments on real datasets obtained from a ride-hailing business show that our method can achieve 10\%-25\% improvement than many state-of-the-art baselines. | Reject |
ICLR.cc/2021/Conference | Systematic generalisation with group invariant predictions | We consider situations where the presence of dominant simpler correlations with the target variable in a training set can cause an SGD-trained neural network to be less reliant on more persistently correlating complex features. When the non-persistent, simpler correlations correspond to non-semantic background factors, a neural network trained on this data can exhibit dramatic failure upon encountering systematic distributional shift, where the correlating background features are recombined with different objects. We perform an empirical study on three synthetic datasets, showing that group invariance methods across inferred partitionings of the training set can lead to significant improvements at such test-time situations. We also suggest a simple invariance penalty, showing with experiments on our setups that it can perform better than alternatives. We find that even without assuming access to any systematically shifted validation sets, one can still find improvements over an ERM-trained reference model. | Accept (Spotlight) |
ICLR.cc/2022/Conference | Zero-Cost Operation Scoring in Differentiable Architecture Search | Differentiable neural architecture search (NAS) has attracted significant attention in recent years due to its ability to quickly discover promising architectures of deep neural networks even in very large search spaces. Despite its success, many differentiable NAS methods lack robustness and may degenerate to trivial architectures with excessive parameter-free operations such as skip connections thus leading to inferior performance. In fact, selecting operations based on the magnitude of architectural parameters was recently proven to be fundamentally wrong, showcasing the need to rethink how operation scoring and selection occurs in differentiable NAS. To this end, we formalize and analyze a fundamental component of differentiable NAS: local "operation scoring" that occurs at each choice of operation.
When comparing existing operation scoring functions, we find that existing methods can be viewed as inexact proxies for accuracy.
We also find that existing methods perform poorly when analyzed empirically on NAS benchmarks. From this perspective, we introduce new training-free proxies to the context of differentiable NAS, and show that we can significantly speed up the search process while improving accuracy on multiple search spaces. We take inspiration from zero-cost proxies that were recently studied in the context of sample-based NAS but shown to degrade significantly for larger search spaces like DARTS. Our novel "perturbation-based zero-cost operation scoring" (Zero-Cost-PT) improves searching time and accuracy compared to the best available differentiable architecture search for many search space sizes, including very large ones. Specifically, we are able improve accuracy compared to the best current method (DARTS-PT) on the DARTS CNN search space while being over 40x faster (total searching time 25 minutes on a single GPU). Our code is available at: https://github.com/avail-upon-acceptance. | Reject |
ICLR.cc/2023/Conference | Regularizing hard examples improves robustness | Recent studies have validated that pruning hard-to-learn examples from training improves the generalization performance of neural networks (NNs). In this study, we investigate this intriguing phenomenon—the negative effect of hard examples on generalization—in adversarial training. Particularly, we theoretically demonstrate that the increase in the difficulty of hard examples in adversarial training is significantly greater than the increase in the difficulty of easy examples. Furthermore, we verify that hard examples are only fitted through memorization of the label in adversarial training and that the memorization of hard examples is attributed to the significant increase in the difficulty of hard examples. We find that the increased difficulty of hard examples brings about the functioning of hard examples as label corrupted data in adversarial training, thereby leading to the memorization of those hard examples and deterioration of the robustness performance. Based upon these observations, we propose a new approach, difficulty proportional label smoothing (DPLS), to mitigate the negative effect of hard examples, thereby improving the adversarial robustness of NNs. Notably, our experimental result indicates that our method can successfully leverage hard examples while circumventing the negative effect. | Withdrawn |
ICLR.cc/2023/Conference | Equivariant Hypergraph Diffusion Neural Operators | Hypergraph neural networks (HNNs) using neural networks to encode hypergraphs provide a promising way to model higher-order relations in data and further solve relevant prediction tasks built upon such higher-order relations. However, higher-order relations in practice contain complex patterns and are often highly irregular. So, it is often challenging to design an HNN that suffices to express those relations while keeping computational efficiency. Inspired by hypergraph diffusion algorithms, this work proposes a new HNN architecture named ED-HNN, which provably approximates any continuous equivariant hypergraph diffusion operators that can model a wide range of higher-order relations. ED-HNN can be implemented efficiently by combining star expansions of hypergraphs with standard message passing neural networks. ED-HNN further shows great superiority in processing heterophilic hypergraphs and constructing deep models. We evaluate ED-HNN for node classification on nine real-world hypergraph datasets. ED-HNN uniformly outperforms the best baselines over these nine datasets and achieves more than 2%$\uparrow$ in prediction accuracy over four datasets therein. Our code is available at: https://github.com/Graph-COM/ED-HNN. | Accept: poster |
ICLR.cc/2021/Conference | Fully Unsupervised Diversity Denoising with Convolutional Variational Autoencoders | Deep Learning based methods have emerged as the indisputable leaders for virtually all image restoration tasks. Especially in the domain of microscopy images, various content-aware image restoration (CARE) approaches are now used to improve the interpretability of acquired data. Naturally, there are limitations to what can be restored in corrupted images, and like for all inverse problems, many potential solutions exist, and one of them must be chosen. Here, we propose DivNoising, a denoising approach based on fully convolutional variational autoencoders (VAEs), overcoming the problem of having to choose a single solution by predicting a whole distribution of denoised images. First we introduce a principled way of formulating the unsupervised denoising problem within the VAE framework by explicitly incorporating imaging noise models into the decoder. Our approach is fully unsupervised, only requiring noisy images and a suitable description of the imaging noise distribution. We show that such a noise model can either be measured, bootstrapped from noisy data, or co-learned during training. If desired, consensus predictions can be inferred from a set of DivNoising predictions, leading to competitive results with other unsupervised methods and, on occasion, even with the supervised state-of-the-art. DivNoising samples from the posterior enable a plethora of useful applications. We are (i) showing denoising results for 13 datasets, (ii) discussing how optical character recognition (OCR) applications can benefit from diverse predictions, and are (iii) demonstrating how instance cell segmentation improves when using diverse DivNoising predictions. | Accept (Poster) |
ICLR.cc/2021/Conference | WeMix: How to Better Utilize Data Augmentation | Data augmentation is a widely used training trick in deep learning to improve the network generalization ability. Despite many encouraging results, several recent studies did point out limitations of the conventional data augmentation scheme in certain scenarios, calling for a better theoretical understanding of data augmentation. In this work, we develop a comprehensive analysis that reveals pros and cons of data augmentation. The main limitation of data augmentation arises from the data bias, i.e. the augmented data distribution can be quite different from the original one. This data bias leads to a suboptimal performance of existing data augmentation methods. To this end, we develop two novel algorithms, termed "AugDrop" and "MixLoss", to correct the data bias in the data augmentation. Our theoretical analysis shows that both algorithms are guaranteed to improve the effect of data augmentation through the bias correction, which is further validated by our empirical studies. Finally, we propose a generic algorithm "WeMix" by combining AugDrop and MixLoss, whose effectiveness is observed from extensive empirical evaluations. | Reject |
ICLR.cc/2021/Conference | Pre-training Text-to-Text Transformers for Concept-centric Common Sense | Pretrained language models (PTLM) have achieved impressive results in a range of natural language understanding (NLU) and generation (NLG) tasks that require a syntactic and semantic understanding of the text. However, current pre-training objectives such as masked token prediction (for BERT-style PTLMs) and masked span infilling (for T5-style PTLMs) do not explicitly model the relational and compositional commonsense knowledge about everyday concepts, which is crucial to many downstream tasks requiring commonsense reasoning. To augment PTLMs with common sense, we propose generative and contrastive objectives as intermediate self-supervised pre-training tasks between general pre-training and downstream task-specific fine-tuning. We also propose a joint training framework to unify generative and contrastive objectives so that these objectives can be more effective.
Our proposed objectives can pack more commonsense knowledge into the parameters of a pre-trained text-to-text transformer without relying on external knowledge bases, yielding better performance on both NLU and NLG tasks. We apply our method on a pre-trained T5 model in an intermediate task transfer learning fashion to train a concept-aware language model (CALM) and experiment with five commonsense benchmarks (four NLU tasks and one NLG task). Experimental results show that CALM outperforms baseline methods by a consistent margin. | Accept (Poster) |
ICLR.cc/2021/Conference | Language-Mediated, Object-Centric Representation Learning | We present Language-mediated, Object-centric Representation Learning (LORL), learning disentangled, object-centric scene representations from vision and language. LORL builds upon recent advances in unsupervised object segmentation, notably MONet and Slot Attention. Just like these algorithms, LORL also learns an object-centric representation by reconstructing the input image. But LORL further learns to associate the learned representations to concepts, i.e., words for object categories, properties, and spatial relationships, from language input. These object-centric concepts derived from language facilitate the learning of object-centric representations. LORL can be integrated with various unsupervised segmentation algorithms that are language-agnostic. Experiments show that LORL consistently improves the performance of MONet and Slot Attention on two datasets via the help of language. We also show that concepts learned by LORL aid downstream tasks such as referential expression interpretation. | Reject |
ICLR.cc/2020/Conference | AHash: A Load-Balanced One Permutation Hash | Minwise Hashing (MinHash) is a fundamental method to compute set similarities and compact high-dimensional data for efficient learning and searching. The bottleneck of MinHash is computing k (usually hundreds) MinHash values. One Permutation Hashing (OPH) only requires one permutation (hash function) to get k MinHash values by dividing elements into k bins. One drawback of OPH is that the load of the bins (the number of elements in a bin) could be unbalanced, which leads to the existence of empty bins and false similarity computation. Several strategies for densification, that is, filling empty bins, have been proposed. However, the densification is just a remedial strategy and cannot eliminate the error incurred by the unbalanced load. Unlike the densification to fill the empty bins after they undesirably occur, our design goal is to balance the load so as to reduce the empty bins in advance. In this paper, we propose a load-balanced hashing, Amortization Hashing (AHash), which can generate as few empty bins as possible. Therefore, AHash is more load-balanced and accurate without hurting runtime efficiency compared with OPH and densification strategies. Our experiments on real datasets validate the claim. All source codes and datasets have been provided as Supplementary Materials and released on GitHub anonymously. | Reject |
ICLR.cc/2019/Conference | Continual Learning via Explicit Structure Learning | Despite recent advances in deep learning, neural networks suffer catastrophic forgetting when tasks are learned sequentially. We propose a conceptually simple and general framework for continual learning, where structure optimization is considered explicitly during learning. We implement this idea by separating the structure and parameter learning. During structure learning, the model optimizes for the best structure for the current task. The model learns when to reuse or modify structure from previous tasks, or create new ones when necessary. The model parameters are then estimated with the optimal structure. Empirically, we found that our approach leads to sensible structures when learning multiple tasks continuously. Additionally, catastrophic forgetting is also largely alleviated from explicit learning of structures. Our method also outperforms all other baselines on the permuted MNIST and split CIFAR datasets in continual learning setting. | Reject |
ICLR.cc/2021/Conference | A Further Study of Unsupervised Pre-training for Transformer Based Speech Recognition | Building a good speech recognition system usually requires large amounts of transcribed data, which is expensive to collect. To tackle this problem, many unsupervised pre-training methods have been proposed. Among these methods, Masked Predictive Coding achieved significant improvements on various speech recognition datasets with BERT-like Masked Reconstruction loss and Transformer backbone. However, many aspects of MPC have not been fully investigated. In this paper, we conduct a further study on MPC and focus on three important aspects: the effect of pre-training data speaking style, its extension on streaming model, and how to better transfer learned knowledge from pre-training stage to downstream tasks. Experiments reveled that pre-training data with matching speaking style is more useful on downstream recognition tasks. A unified training objective with APC and MPC provided 8.46% relative error reduction on streaming model trained on HKUST. Also, the combination of target data adaption and layer-wise discriminative training helped the knowledge transfer of MPC, which achieved 3.99% relative error reduction on AISHELL over a strong baseline. | Desk_Rejected |
ICLR.cc/2018/Conference | DEEPCAST : UNIVERSAL TIME-SERIES FORECASTER | Reliable and accurate time-series forecasting is critical in many fields including energy, finance, and manufacturing. Many time-series tasks, however, suffer from a limited amount of training data (i.e., the cold start problem) resulting in poor forecasting performance. Recently, convolutional neural networks (CNNs) have shown outstanding image classification performance even on tasks with small-scale training sets. The performance can be attributed to transfer learning through CNNs’ ability to learn rich mid-level image representations. However, no prior work exists on general transfer learning for time-series forecasting. In this paper, motivated by recent success of transfer learning in CNN model and image-related tasks, we for the first time show how time-series representations learned with Long Short Term Memory (LSTM) on large-scale datasets can be efficiently transferred to other time-series forecasting tasks with limited amount of training data. We also validate that despite differences in time-series statistics and tasks in the datasets, the transferred representation leads to significantly improved forecasting results outperforming majority of the best time-series methods on the public M3 and other datasets. Our online universal forecasting tool, DeepCast, will leverage transfer learning to provide accurate forecasts for a diverse set of time series where classical methods were computationally infeasible or inapplicable due to short training history. | Withdrawn |
ICLR.cc/2021/Conference | Learning to Represent Action Values as a Hypergraph on the Action Vertices | Action-value estimation is a critical component of many reinforcement learning (RL) methods whereby sample complexity relies heavily on how fast a good estimator for action value can be learned. By viewing this problem through the lens of representation learning, good representations of both state and action can facilitate action-value estimation. While advances in deep learning have seamlessly driven progress in learning state representations, given the specificity of the notion of agency to RL, little attention has been paid to learning action representations. We conjecture that leveraging the combinatorial structure of multi-dimensional action spaces is a key ingredient for learning good representations of action. To test this, we set forth the action hypergraph networks framework---a class of functions for learning action representations in multi-dimensional discrete action spaces with a structural inductive bias. Using this framework we realise an agent class based on a combination with deep Q-networks, which we dub hypergraph Q-networks. We show the effectiveness of our approach on a myriad of domains: illustrative prediction problems under minimal confounding effects, Atari 2600 games, and discretised physical control benchmarks. | Accept (Poster) |
ICLR.cc/2022/Conference | BWCP: Probabilistic Learning-to-Prune Channels for ConvNets via Batch Whitening | This work presents a probabilistic channel pruning method to accelerate Convolutional Neural Networks (CNNs). Previous pruning methods often zero out unimportant channels in training in a deterministic manner, which reduces CNN's learning capacity and results in suboptimal performance. To address this problem, we develop a probability-based pruning algorithm, called batch whitening channel pruning (BWCP), which can stochastically discard unimportant channels by modeling the probability of a channel being activated. BWCP has several merits. (1) It simultaneously trains and prunes CNNs from scratch in a probabilistic way, exploring larger network space than deterministic methods. (2) BWCP is empowered by the proposed batch whitening tool, which is able to empirically and theoretically increase the activation probability of useful channels while reducing the probability of unimportant channels without adding any extra parameters and computational cost in inference. (3) Extensive experiments on CIFAR-10, CIFAR-100, and ImageNet with various network architectures show that BWCP outperforms its counterparts by achieving better accuracy given limited computational budgets. For example, ResNet50 pruned by BWCP has only 0.58% Top-1 accuracy drop on ImageNet, while reducing 42.9% FLOPs of the plain ResNet50. | Reject |
ICLR.cc/2023/Conference | A Time Series is Worth 64 Words: Long-term Forecasting with Transformers | We propose an efficient design of Transformer-based models for multivariate time series forecasting and self-supervised representation learning. It is based on two key components: (i) segmentation of time series into subseries-level patches which are served as input tokens to Transformer; (ii) channel-independence where each channel contains a single univariate time series that shares the same embedding and Transformer weights across all the series. Patching design naturally has three-fold benefit: local semantic information is retained in the embedding; computation and memory usage of the attention maps are quadratically reduced given the same look-back window; and the model can attend longer history. Our channel-independent patch time series Transformer (PatchTST) can improve the long-term forecasting accuracy significantly when compared with that of SOTA Transformer-based models. We also apply our model to self-supervised pre-training tasks and attain excellent fine-tuning performance, which outperforms supervised training on large datasets. Transferring of masked pre-training performed on one dataset to other datasets also produces SOTA forecasting accuracy. | Accept: poster |
ICLR.cc/2021/Conference | HW-NAS-Bench: Hardware-Aware Neural Architecture Search Benchmark | HardWare-aware Neural Architecture Search (HW-NAS) has recently gained tremendous attention by automating the design of deep neural networks deployed in more resource-constrained daily life devices. Despite its promising performance, developing optimal HW-NAS solutions can be prohibitively challenging as it requires cross-disciplinary knowledge in the algorithm, micro-architecture, and device-specific compilation. First, to determine the hardware-cost to be incorporated into the NAS process, existing works mostly adopt either pre-collected hardware-cost look-up tables or device-specific hardware-cost models. The former can be time-consuming due to the required knowledge of the device’s compilation method and how to set up the measurement pipeline, while building the latter is often a barrier for non-hardware experts like NAS researchers. Both of them limit the development of HW-NAS innovations and impose a barrier-to-entry to non-hardware experts. Second, similar to generic NAS, it can be notoriously difficult to benchmark HW-NAS algorithms due to their significant required computational resources and the differences in adopted search spaces, hyperparameters, and hardware devices. To this end, we develop HW-NAS-Bench, the first public dataset for HW-NAS research which aims to democratize HW-NAS research to non-hardware experts and make HW-NAS research more reproducible and accessible. To design HW-NAS-Bench, we carefully collected the measured/estimated hardware performance (e.g., energy cost and latency) of all the networks in the search spaces of both NAS-Bench-201 and FBNet, on six hardware devices that fall into three categories (i.e., commercial edge devices, FPGA, and ASIC). Furthermore, we provide a comprehensive analysis of the collected measurements in HW-NAS-Bench to provide insights for HW-NAS research. Finally, we demonstrate exemplary user cases to (1) show that HW-NAS-Bench allows non-hardware experts to perform HW-NAS by simply querying our pre-measured dataset and (2) verify that dedicated device-specific HW-NAS can indeed lead to optimal accuracy-cost trade-offs. The codes and all collected data are available at https://github.com/RICE-EIC/HW-NAS-Bench. | Accept (Spotlight) |
ICLR.cc/2023/Conference | MESSAGENET: MESSAGE CLASSIFICATION USING NATURAL LANGUAGE PROCESSING AND META-DATA | In this paper we propose a new Deep Learning (DL) approach for message classification. Our method
is based on the state-of-the-art Natural Language Processing (NLP) building blocks, combined
with a novel technique for infusing the meta-data input that is typically available in messages
such as the sender information, timestamps, attached image, audio, affiliations, and more. As we
demonstrate throughout the paper, going beyond the mere text by leveraging all available channels
in the message, could yield an improved representation and higher classification accuracy. To
achieve message representation, each type of input is processed in a dedicated block in the neural
network architecture that is suitable for the data type. Such an implementation enables training all
blocks together simultaneously, and forming cross channels features in the network. We show in the
Experiments Section that in some cases, message’s meta-data holds an additional information that
cannot be extracted just from the text, and when using this information we achieve better performance.
Furthermore, we demonstrate that our multi-modality block approach outperforms other approaches
for injecting the meta data to the the text classifier. | Reject |