Learning deep representations by mutual information estimation and maximization

Abstract

In this work, we perform unsupervised learning of representations by maximizing mutual information between an input and the output of a deep neural network encoder. Importantly, we show that structure matters: incorporating knowledge about locality of the input to the objective can greatly influence a representation's suitability for downstream tasks. We further control characteristics of the representation by matching to a prior distribution adversarially. Our method, which we call Deep InfoMax (DIM), outperforms a number of popular unsupervised learning methods and competes with fully-supervised learning on several classification tasks. DIM opens new avenues for unsupervised learning of representations and is an important step towards flexible formulations of representation-learning objectives for specific end-goals.

Accepted as an oral presentation at the International Conference for Learning Representations (ICLR), 2019

Cited by in corpus (426)

• A Simple Framework for Contrastive Learning of Visual Representations
• Shortcut Learning in Deep Neural Networks
• Improved Baselines with Momentum Contrastive Learning
• Unsupervised Learning of Visual Features by Contrasting Cluster Assignments
• Pre-trained Models for Natural Language Processing: A Survey
• Graph Contrastive Learning with Adaptive Augmentation
• Momentum Contrast for Unsupervised Visual Representation Learning
• BEiT: BERT Pre-Training of Image Transformers
• Data-Efficient Image Recognition with Contrastive Predictive Coding
• Graph Contrastive Learning with Augmentations
• Contrastive Representation Learning: A Framework and Review
• Learning Representations by Maximizing Mutual Information Across Views
• What Makes for Good Views for Contrastive Learning?
• Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges
• Contrastive Multiview Coding
• Graph Self-Supervised Learning: A Survey
• Anomaly Detection on Attributed Networks via Contrastive Self-Supervised Learning
• Prototypical Contrastive Learning of Unsupervised Representations
• Big Self-Supervised Models are Strong Semi-Supervised Learners
• Contrastive Learning for Representation Degeneration Problem in Sequential Recommendation
• A Review on Deep Learning in UAV Remote Sensing
• Exploring Simple Siamese Representation Learning
• Deep Graph Contrastive Representation Learning
• Self-paced Contrastive Learning with Hybrid Memory for Domain Adaptive Object Re-ID
• Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty
• Towards artificial general intelligence via a multimodal foundation model
• On Mutual Information Maximization for Representation Learning
• Self-supervised Co-training for Video Representation Learning
• Hard Negative Mixing for Contrastive Learning
• Out-of-Distribution Generalization via Risk Extrapolation (REx)
• A Primer on Motion Capture with Deep Learning: Principles, Pitfalls and Perspectives
• Learning Video Representations using Contrastive Bidirectional Transformer
• A survey on Semi-, Self- and Unsupervised Learning for Image Classification
• CURL: Contrastive Unsupervised Representations for Reinforcement Learning
• Self-Supervised MultiModal Versatile Networks
• Momentum Contrastive Learning for Few-Shot COVID-19 Diagnosis from Chest CT Images
• Multi-view Graph Contrastive Representation Learning for Drug-Drug Interaction Prediction
• Multi-View Graph Convolutional Network for Multimedia Recommendation
• Understanding Contrastive Representation Learning through Alignment and Uniformity on the Hypersphere
• A Review-aware Graph Contrastive Learning Framework for Recommendation
• A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark
• Pre-training Molecular Graph Representation with 3D Geometry
• Invertible Residual Networks
• HDMI: High-order Deep Multiplex Infomax
• Knowledge Graph Contrastive Learning Based on Relation-Symmetrical Structure
• InfoGraph: Unsupervised and Semi-supervised Graph-Level Representation Learning via Mutual Information Maximization
• Adversarial Graph Augmentation to Improve Graph Contrastive Learning
• Social Recommendation with Self-Supervised Metagraph Informax Network
• Space-Time Correspondence as a Contrastive Random Walk
• Contrastive Cross-Domain Sequential Recommendation
• CrossCBR: Cross-view Contrastive Learning for Bundle Recommendation
• Consistent Multiple Graph Embedding for Multi-View Clustering
• Self-supervised remote sensing feature learning: Learning Paradigms, Challenges, and Future Works
• The Conditional Entropy Bottleneck
• Knowledge-Preserving Incremental Social Event Detection via Heterogeneous GNNs
• Graph Information Bottleneck
• Model Adaptation: Historical Contrastive Learning for Unsupervised Domain Adaptation without Source Data
• PointContrast: Unsupervised Pre-training for 3D Point Cloud Understanding
• 3D Infomax improves GNNs for Molecular Property Prediction
• Efficient Self-supervised Vision Transformers for Representation Learning
• Learning Robust Representations via Multi-View Information Bottleneck
• A Mutual Information Maximization Perspective of Language Representation Learning
• Putting An End to End-to-End: Gradient-Isolated Learning of Representations
• Selfie: Self-supervised Pretraining for Image Embedding
• InfoXLM: An Information-Theoretic Framework for Cross-Lingual Language Model Pre-Training
• A Practical Contrastive Learning Framework for Single-Image Super-Resolution
• A Good Image Generator Is What You Need for High-Resolution Video Synthesis
• Self-supervised learning through the eyes of a child
• Heterogeneous Deep Graph Infomax
• Contrastive Representation Distillation
• Intriguing Properties of Contrastive Losses
• Self-Supervised Learning with Data Augmentations Provably Isolates Content from Style
• A Systematic Survey of Regularization and Normalization in GANs
• Video Representation Learning with Visual Tempo Consistency
• InfoBERT: Improving Robustness of Language Models from An Information Theoretic Perspective
• Formal Limitations on the Measurement of Mutual Information
• Supervised Contrastive Learning for Pre-trained Language Model Fine-tuning
• Cluster Contrast for Unsupervised Person Re-Identification
• Near-Optimal Representation Learning for Hierarchical Reinforcement Learning
• Dual Space Graph Contrastive Learning
• A Survey of Deep Learning for Scientific Discovery
• Excessive Invariance Causes Adversarial Vulnerability
• A Review of Predictive and Contrastive Self-supervised Learning for Medical Images
• CoMIR: Contrastive Multimodal Image Representation for Registration
• i-Mix: A Domain-Agnostic Strategy for Contrastive Representation Learning
• ReduNet: A White-box Deep Network from the Principle of Maximizing Rate Reduction
• Learning Dynamic Belief Graphs to Generalize on Text-Based Games
• HERO: Hierarchical Encoder for Video+Language Omni-representation Pre-training
• An information-theoretic perspective on intrinsic motivation in reinforcement learning: a survey
• Disentangling Disentanglement in Variational Autoencoders
• FairFil: Contrastive Neural Debiasing Method for Pretrained Text Encoders
• Graph Pooling via Coarsened Graph Infomax
• Jointly Contrastive Representation Learning on Road Network and Trajectory
• Graph Clustering with Graph Neural Networks
• Unsupervised Deep Representation Learning and Few-Shot Classification of PolSAR Images
• From Canonical Correlation Analysis to Self-supervised Graph Neural Networks
• Transductive Information Maximization For Few-Shot Learning
• Data-Efficient Reinforcement Learning with Self-Predictive Representations
• Predicting What You Already Know Helps: Provable Self-Supervised Learning
• CCGL: Contrastive Cascade Graph Learning
• Active Contrastive Learning of Audio-Visual Video Representations
• Contrastive Learning for Recommender System
• Physically-Constrained Transfer Learning through Shared Abundance Space for Hyperspectral Image Classification
• Automatic Shortcut Removal for Self-Supervised Representation Learning
• Whitening for Self-Supervised Representation Learning
• Provable Guarantees for Self-Supervised Deep Learning with Spectral Contrastive Loss
• Representation Learning via Invariant Causal Mechanisms
• Contrastive Learning of Single-Cell Phenotypic Representations for Treatment Classification
• Nonlinear ICA Using Auxiliary Variables and Generalized Contrastive Learning
• CGC: Contrastive Graph Clustering for Community Detection and Tracking
• Contrastive Knowledge Graph Error Detection
• Multi-Label Contrastive Learning for Abstract Visual Reasoning
• Watching the World Go By: Representation Learning from Unlabeled Videos
• PGL: Prior-Guided Local Self-supervised Learning for 3D Medical Image Segmentation
• Parameter Efficient Multimodal Transformers for Video Representation Learning
• Rethinking Softmax with Cross-Entropy: Neural Network Classifier as Mutual Information Estimator
• Contrastive Learning with Stronger Augmentations
• Stacked Capsule Autoencoders
• Deep Verifier Networks: Verification of Deep Discriminative Models with Deep Generative Models
• Self-Supervised Hypergraph Convolutional Networks for Session-based Recommendation
• Learning Spatiotemporal Features via Video and Text Pair Discrimination
• Pretraining Representations for Data-Efficient Reinforcement Learning
• L_DMI: An Information-theoretic Noise-robust Loss Function
• Contrastive Learning of Musical Representations
• Self-Supervised Multi-Channel Hypergraph Convolutional Network for Social Recommendation
• Understanding the Limitations of Variational Mutual Information Estimators
• The Lottery Tickets Hypothesis for Supervised and Self-supervised Pre-training in Computer Vision Models
• Inherit with Distillation and Evolve with Contrast: Exploring Class Incremental Semantic Segmentation Without Exemplar Memory
• Large-Scale Representation Learning on Graphs via Bootstrapping
• Multi-Format Contrastive Learning of Audio Representations
• Semi-Supervised Contrastive Learning with Generalized Contrastive Loss and Its Application to Speaker Recognition
• MST: Masked Self-Supervised Transformer for Visual Representation
• Deep Semi-Supervised Anomaly Detection
• FSCE: Few-Shot Object Detection via Contrastive Proposal Encoding
• Towards Robust Graph Contrastive Learning
• Contrastive Learning Inverts the Data Generating Process
• CoDA: Contrast-enhanced and Diversity-promoting Data Augmentation for Natural Language Understanding
• Contrastive Multi-Level Graph Neural Networks for Session-based Recommendation
• SelfMatch: Combining Contrastive Self-Supervision and Consistency for Semi-Supervised Learning
• Arbitrary Talking Face Generation via Attentional Audio-Visual Coherence Learning
• Self-Supervised Relational Reasoning for Representation Learning
• Low-Resolution Object Recognition with Cross-Resolution Relational Contrastive Distillation
• MedAug: Contrastive learning leveraging patient metadata improves representations for chest X-ray interpretation
• Transfer Learning of Graph Neural Networks with Ego-graph Information Maximization
• The Ingredients of Real-World Robotic Reinforcement Learning
• Unsupervised State Representation Learning in Atari
• A Text-guided Protein Design Framework
• Incremental False Negative Detection for Contrastive Learning
• Detecting Arbitrary Order Beneficial Feature Interactions for Recommender Systems
• Is Image-to-Image Translation the Panacea for Multimodal Image Registration? A Comparative Study
• SUGAR: Subgraph Neural Network with Reinforcement Pooling and Self-Supervised Mutual Information Mechanism
• Evaluating Contrastive Models for Instance-based Image Retrieval
• Revisiting Locally Supervised Learning: an Alternative to End-to-end Training
• Learning from 2D: Contrastive Pixel-to-Point Knowledge Transfer for 3D Pretraining
• Self-Supervised Time Series Representation Learning by Inter-Intra Relational Reasoning
• Semantic Information G Theory and Logical Bayesian Inference for Machine Learning
• Deep Reinforcement and InfoMax Learning
• Predictive Information Accelerates Learning in RL
• Neural Transformation Learning for Deep Anomaly Detection Beyond Images
• Boosting Semi-supervised Image Segmentation with Global and Local Mutual Information Regularization
• Easy Transfer Learning By Exploiting Intra-domain Structures
• Joint Contrastive Learning for Unsupervised Domain Adaptation
• Unsupervised Semantic Aggregation and Deformable Template Matching for Semi-Supervised Learning
• Towards the Generalization of Contrastive Self-Supervised Learning
• Towards Graph Self-Supervised Learning with Contrastive Adjusted Zooming
• CLDG: Contrastive Learning on Dynamic Graphs
• Relational Self-Supervised Learning on Graphs
• Image Clustering using an Augmented Generative Adversarial Network and Information Maximization
• Neural Approximate Sufficient Statistics for Implicit Models
• Unsupervised Attributed Multiplex Network Embedding
• Self-supervised EEG Representation Learning for Automatic Sleep Staging
• Stochastic Mutual Information Gradient Estimation for Dimensionality Reduction Networks
• FedDCSR: Federated Cross-domain Sequential Recommendation via Disentangled Representation Learning
• Learning Optimal Representations with the Decodable Information Bottleneck
• A Large-Scale Study on Unsupervised Spatiotemporal Representation Learning
• Multi-View Class Incremental Learning
• Estimating Information Flow in Deep Neural Networks
• Conditional Mutual Information Neural Estimator
• Enhancing Unsupervised Video Representation Learning by Decoupling the Scene and the Motion
• DSTSA-GCN: Advancing Skeleton-Based Gesture Recognition with Semantic-Aware Spatio-Temporal Topology Modeling
• Deep Comprehensive Correlation Mining for Image Clustering
• Augmented Skeleton Based Contrastive Action Learning with Momentum LSTM for Unsupervised Action Recognition
• Utilizing Edge Features in Graph Neural Networks via Variational Information Maximization
• Contrastive learning, multi-view redundancy, and linear models
• Multimodal Contrastive Training for Visual Representation Learning
• Integrating Contrastive Learning with Dynamic Models for Reinforcement Learning from Images
• Exploring Data-Efficient 3D Scene Understanding with Contrastive Scene Contexts
• An Asymmetric Contrastive Loss for Handling Imbalanced Datasets
• Motif-Driven Contrastive Learning of Graph Representations
• Hierarchical Aggregations for High-Dimensional Multiplex Graph Embedding
• SoundCLR: Contrastive Learning of Representations For Improved Environmental Sound Classification
• Discovering and Explaining the Representation Bottleneck of DNNs
• Momentum Contrastive Voxel-wise Representation Learning for Semi-supervised Volumetric Medical Image Segmentation
• When Does Self-supervision Improve Few-shot Learning?
• How to Train Your Energy-Based Model for Regression
• USER: Unsupervised Structural Entropy-based Robust Graph Neural Network
• Joint Learning of Localized Representations from Medical Images and Reports
• Universum-inspired Supervised Contrastive Learning
• CAPT: Contrastive Pre-Training for Learning Denoised Sequence Representations
• Contrastive Predictive Coding for Anomaly Detection
• Contrastively Disentangled Sequential Variational Autoencoder
• A Note on Connecting Barlow Twins with Negative-Sample-Free Contrastive Learning
• A Review of Artificial Intelligence Technologies for Early Prediction of Alzheimer's Disease
• Understanding and Improving Robustness of Vision Transformers through Patch-based Negative Augmentation
• Unsupervised Feature Learning by Cross-Level Instance-Group Discrimination
• Modeling Multiple Views via Implicitly Preserving Global Consistency and Local Complementarity
• Frustratingly Easy Transferability Estimation
• Contrastive Learning for Cold-Start Recommendation
• Open-Set Hypothesis Transfer with Semantic Consistency
• Global-Local Bidirectional Reasoning for Unsupervised Representation Learning of 3D Point Clouds
• Neural Entropic Estimation: A faster path to mutual information estimation
• Contrastive and Generative Graph Convolutional Networks for Graph-based Semi-Supervised Learning
• Farewell to Mutual Information: Variational Distillation for Cross-Modal Person Re-Identification
• Joint Contrastive Learning with Infinite Possibilities
• Unsupervised Learning from Video with Deep Neural Embeddings
• Self-supervised Learning from a Multi-view Perspective
• Doubly Contrastive Deep Clustering
• Contrastive Learning for Debiased Candidate Generation in Large-Scale Recommender Systems
• Tight Mutual Information Estimation With Contrastive Fenchel-Legendre Optimization
• One Size Does Not Fit All: Generating and Evaluating Variable Number of Keyphrases
• Iterative Graph Self-Distillation
• M2IOSR: Maximal Mutual Information Open Set Recognition
• Relating by Contrasting: A Data-efficient Framework for Multimodal Generative Models
• Category Contrast for Unsupervised Domain Adaptation in Visual Tasks
• SCARF: Self-Supervised Contrastive Learning using Random Feature Corruption
• The Spatially-Correlative Loss for Various Image Translation Tasks
• CO2: Consistent Contrast for Unsupervised Visual Representation Learning
• Fairness-Aware Node Representation Learning
• Learning Disentangled Representations via Mutual Information Estimation
• Improving Transformation Invariance in Contrastive Representation Learning
• Probing the State of the Art: A Critical Look at Visual Representation Evaluation
• Towards Domain-Agnostic Contrastive Learning
• Simpler, Faster, Stronger: Breaking The log-K Curse On Contrastive Learners With FlatNCE
• Zero-Shot Learning from scratch (ZFS): leveraging local compositional representations
• SCE: Scalable Network Embedding from Sparsest Cut
• DimCL: Dimensional Contrastive Learning For Improving Self-Supervised Learning
• Multi-label Contrastive Predictive Coding
• Heterogeneous Social Event Detection via Hyperbolic Graph Representations
• PointSmile: Point Self-supervised Learning via Curriculum Mutual Information
• Learning Adversarially Robust Representations via Worst-Case Mutual Information Maximization
• Drop-Bottleneck: Learning Discrete Compressed Representation for Noise-Robust Exploration
• PlayVirtual: Augmenting Cycle-Consistent Virtual Trajectories for Reinforcement Learning
• ESAD: End-to-end Deep Semi-supervised Anomaly Detection
• Data-Efficient Pretraining via Contrastive Self-Supervision
• Pretext-Contrastive Learning: Toward Good Practices in Self-supervised Video Representation Leaning
• flexgrid2vec: Learning Efficient Visual Representations Vectors
• Improving Weakly Supervised Visual Grounding by Contrastive Knowledge Distillation
• Self-Supervised Learning Aided Class-Incremental Lifelong Learning
• Learning Discrete Structured Representations by Adversarially Maximizing Mutual Information
• Refining Pseudo Labels with Clustering Consensus over Generations for Unsupervised Object Re-identification
• Mutual Information Gradient Estimation for Representation Learning
• Understanding Hyperbolic Metric Learning through Hard Negative Sampling
• Language-Driven Image Style Transfer
• Renormalized Mutual Information for Artificial Scientific Discovery
• QAInfomax: Learning Robust Question Answering System by Mutual Information Maximization
• Unsupervised Object-Level Representation Learning from Scene Images
• CSNNs: Unsupervised, Backpropagation-free Convolutional Neural Networks for Representation Learning
• A Realistic Evaluation of Semi-Supervised Learning for Fine-Grained Classification
• Cross-Trajectory Representation Learning for Zero-Shot Generalization in RL
• Neuron Campaign for Initialization Guided by Information Bottleneck Theory
• DICE: Diversity in Deep Ensembles via Conditional Redundancy Adversarial Estimation
• Privacy-Preserving Representation Learning on Graphs: A Mutual Information Perspective
• Reviewing Evolution of Learning Functions and Semantic Information Measures for Understanding Deep Learning
• Bridging Imagination and Reality for Model-Based Deep Reinforcement Learning
• Quantifying and Mitigating Privacy Risks of Contrastive Learning
• Skill Disentanglement for Imitation Learning from Suboptimal Demonstrations
• Self-Supervised Multi-View Learning via Auto-Encoding 3D Transformations
• Mutual Information Maximization for Effective Lip Reading
• Deep Deterministic Information Bottleneck with Matrix-based Entropy Functional
• Auxiliary Tasks Speed Up Learning PointGoal Navigation
• Representation Learning via Consistent Assignment of Views to Clusters
• Neural Methods for Point-wise Dependency Estimation
• CoKe: Localized Contrastive Learning for Robust Keypoint Detection
• Calibrating and Improving Graph Contrastive Learning
• Mitigating Spurious Correlations for Self-supervised Recommendation
• CUCL: Codebook for Unsupervised Continual Learning
• Variational Autoencoders for Opponent Modeling in Multi-Agent Systems
• DAGNN: Demand-aware Graph Neural Networks for Session-based Recommendation
• Direct Feedback Alignment Scales to Modern Deep Learning Tasks and Architectures
• Telescoping Density-Ratio Estimation
• CLCC: Contrastive Learning for Color Constancy
• Emergence of Human-Like Attention in Self-Supervised Vision Transformers: an eye-tracking study
• Hybrid Generative-Contrastive Representation Learning
• Biphasic Learning of GANs for High-Resolution Image-to-Image Translation
• DHOG: Deep Hierarchical Object Grouping
• Feature Adaptation of Pre-Trained Language Models across Languages and Domains with Robust Self-Training
• Distilling Audio-Visual Knowledge by Compositional Contrastive Learning
• Sampling and Recovery of Graph Signals based on Graph Neural Networks
• Can Pretext-Based Self-Supervised Learning Be Boosted by Downstream Data? A Theoretical Analysis
• Cross-Modal Information Maximization for Medical Imaging: CMIM
• On the Estimation of Information Measures of Continuous Distributions
• Focus of Attention Improves Information Transfer in Visual Features
• InfoFair: Information-Theoretic Intersectional Fairness
• Self-supervised Tumor Segmentation through Layer Decomposition
• Improving the Learning of Multi-column Convolutional Neural Network for Crowd Counting
• Unsupervised Detection of Distinctive Regions on 3D Shapes
• Predictive Coding for Boosting Deep Reinforcement Learning with Sparse Rewards
• Graph Cross Networks with Vertex Infomax Pooling
• Generative Hierarchical Features from Synthesizing Images
• AutoLTS: Automating Cycling Stress Assessment via Contrastive Learning and Spatial Post-processing
• On Feature Decorrelation in Self-Supervised Learning
• A Semantic Generalization of Shannon's Information Theory and Applications
• Function Contrastive Learning of Transferable Meta-Representations
• Improving Contrastive Learning by Visualizing Feature Transformation
• Causality-inspired Single-source Domain Generalization for Medical Image Segmentation
• Abstraction Mechanisms Predict Generalization in Deep Neural Networks
• Tuning-Free Disentanglement via Projection
• Gradient-based Bayesian Experimental Design for Implicit Models using Mutual Information Lower Bounds
• Conditional Mutual information-based Contrastive Loss for Financial Time Series Forecasting
• Hierarchical Self-Supervised Learning for Medical Image Segmentation Based on Multi-Domain Data Aggregation
• Interventional Video Grounding with Dual Contrastive Learning
• S2-BNN: Bridging the Gap Between Self-Supervised Real and 1-bit Neural Networks via Guided Distribution Calibration
• Mutual Information State Intrinsic Control
• DRIBO: Robust Deep Reinforcement Learning via Multi-View Information Bottleneck
• When Contrastive Learning Meets Active Learning: A Novel Graph Active Learning Paradigm with Self-Supervision
• Contrastive Attraction and Contrastive Repulsion for Representation Learning
• Learning Visual Representations for Transfer Learning by Suppressing Texture
• ProtoInfoMax: Prototypical Networks with Mutual Information Maximization for Out-of-Domain Detection
• Evaluating Modules in Graph Contrastive Learning
• Multi-Sample based Contrastive Loss for Top-k Recommendation
• LSMI-Sinkhorn: Semi-supervised Mutual Information Estimation with Optimal Transport
• Unsupervised Part Discovery from Contrastive Reconstruction
• An Empirical Study of Graph Contrastive Learning
• Multimodal Image-to-Image Translation via Mutual Information Estimation and Maximization
• Pri3D: Can 3D Priors Help 2D Representation Learning?
• Multimodal Representation Learning via Maximization of Local Mutual Information
• Learning to Compare Relation: Semantic Alignment for Few-Shot Learning
• Representation Learning for Remote Sensing: An Unsupervised Sensor Fusion Approach
• Memory Augmented Multi-Instance Contrastive Predictive Coding for Sequential Recommendation
• ANL: Anti-Noise Learning for Cross-Domain Person Re-Identification
• An end-to-end approach for the verification problem: learning the right distance
• Tasting the cake: evaluating self-supervised generalization on out-of-distribution multimodal MRI data
• The Information Geometry of Unsupervised Reinforcement Learning
• Mutual Information-based State-Control for Intrinsically Motivated Reinforcement Learning
• Learning View-Disentangled Human Pose Representation by Contrastive Cross-View Mutual Information Maximization
• MUSCLE: Strengthening Semi-Supervised Learning Via Concurrent Unsupervised Learning Using Mutual Information Maximization
• MIM: Mutual Information Machine
• Integrating Categorical Semantics into Unsupervised Domain Translation
• Addressing Feature Suppression in Unsupervised Visual Representations
• Deep Contrastive Multiview Network Embedding
• Compressive Visual Representations
• Don't miss the Mismatch: Investigating the Objective Function Mismatch for Unsupervised Representation Learning
• Unsupervised Image Classification for Deep Representation Learning
• Contrastive Learning of Global-Local Video Representations
• MISS: Multi-Interest Self-Supervised Learning Framework for Click-Through Rate Prediction
• Permutation-Invariant Variational Autoencoder for Graph-Level Representation Learning
• Smart Home Appliances: Chat with Your Fridge
• Matching Text with Deep Mutual Information Estimation
• Multi-view Contrastive Learning for Online Knowledge Distillation
• Shuffle & Divide: Contrastive Learning for Long Text
• Unsupervised Natural Language Inference via Decoupled Multimodal Contrastive Learning
• PoseContrast: Class-Agnostic Object Viewpoint Estimation in the Wild with Pose-Aware Contrastive Learning
• On the difficulty of learning and predicting the long-term dynamics of bouncing objects
• Unsupervised Path Representation Learning with Curriculum Negative Sampling
• KLDivNet: An unsupervised neural network for multi-modality image registration
• PanRep: Graph neural networks for extracting universal node embeddings in heterogeneous graphs
• Contrastive Learning of Relative Position Regression for One-Shot Object Localization in 3D Medical Images
• Learning Multi-Granular Hypergraphs for Video-Based Person Re-Identification
• Maximizing Mutual Information Across Feature and Topology Views for Learning Graph Representations
• Mutual Information Regularized Identity-aware Facial ExpressionRecognition in Compressed Video
• Representation Learning for Sequence Data with Deep Autoencoding Predictive Components
• Point Discriminative Learning for Data-efficient 3D Point Cloud Analysis
• Information Theory-Guided Heuristic Progressive Multi-View Coding
• Interest-Behaviour Multiplicative Network for Resource-limited Recommendation
• ExCon: Explanation-driven Supervised Contrastive Learning for Image Classification
• Self-Distilled Self-Supervised Representation Learning
• Neural Bayes: A Generic Parameterization Method for Unsupervised Representation Learning
• Obstacle Identification and Ellipsoidal Decomposition for Fast Motion Planning in Unknown Dynamic Environments
• K-Shot Contrastive Learning of Visual Features with Multiple Instance Augmentations
• Intrinsic and Extrinsic Factor Disentanglement for Recommendation in Various Context Scenarios
• ContraCluster: Learning to Classify without Labels by Contrastive Self-Supervision and Prototype-Based Semi-Supervision
• On the robustness of self-supervised representations for multi-view object classification
• Contrastive Adaptive Propagation Graph Neural Networks for Efficient Graph Learning
• CLAWS: Contrastive Learning with hard Attention and Weak Supervision
• Contrastive Video Representation Learning via Adversarial Perturbations
• On Contrastive Representations of Stochastic Processes
• Decoupling Global and Local Representations via Invertible Generative Flows
• Toward Interpretability of Dual-Encoder Models for Dialogue Response Suggestions
• Unsupervised Hierarchical Graph Representation Learning by Mutual Information Maximization
• Efficient Person Search: An Anchor-Free Approach
• Self-Supervised Video Representation Learning by Video Incoherence Detection
• Learning Robust Representation for Clustering through Locality Preserving Variational Discriminative Network
• Learning Deep Representations by Mutual Information for Person Re-identification
• Graph Embedding Using Infomax for ASD Classification and Brain Functional Difference Detection
• Blocked and Hierarchical Disentangled Representation From Information Theory Perspective
• Non-Local Representation based Mutual Affine-Transfer Network for Photorealistic Stylization
• PointShuffleNet: Learning Non-Euclidean Features with Homotopy Equivalence and Mutual Information
• Better Long-Range Dependency By Bootstrapping A Mutual Information Regularizer
• MI^2GAN: Generative Adversarial Network for Medical Image Domain Adaptation using Mutual Information Constraint
• Self-supervised SAR-optical Data Fusion and Land-cover Mapping using Sentinel-1/-2 Images
• In Defense of LSTMs for Addressing Multiple Instance Learning Problems
• Unsupervised Abstract Reasoning for Raven's Problem Matrices
• On Compositions of Transformations in Contrastive Self-Supervised Learning
• High Mutual Information in Representation Learning with Symmetric Variational Inference
• Guess First to Enable Better Compression and Adversarial Robustness
• Self-supervised Learning with Local Attention-Aware Feature
• Leveraging Hidden Structure in Self-Supervised Learning
• Learning Transferable Concepts in Deep Reinforcement Learning
• Channel Recurrent Attention Networks for Video Pedestrian Retrieval
• AI Progress in Skin Lesion Analysis
• Neural Network Classifier as Mutual Information Evaluator
• Contrastive Separative Coding for Self-supervised Representation Learning
• Synthesizing Photorealistic Images with Deep Generative Learning
• Multi-Pretext Attention Network for Few-shot Learning with Self-supervision
• Analyzing the Hidden Activations of Deep Policy Networks: Why Representation Matters
• Robust Robotic Control from Pixels using Contrastive Recurrent State-Space Models
• Explaining Representation by Mutual Information
• On Equivariant and Invariant Learning of Object Landmark Representations
• PreViTS: Contrastive Pretraining with Video Tracking Supervision
• Think Global, Act Local: Relating DNN generalisation and node-level SNR
• MC-SSL0.0: Towards Multi-Concept Self-Supervised Learning
• Contrastive Representation Learning with Trainable Augmentation Channel
• Moreau-Yosida $f$-divergences
• Label-Based Diversity Measure Among Hidden Units of Deep Neural Networks: A Regularization Method
• Deep Bregman Divergence for Contrastive Learning of Visual Representations
• $C^3$: Compositional Counterfactual Contrastive Learning for Video-grounded Dialogues
• Characterizing and Improving the Robustness of Self-Supervised Learning through Background Augmentations
• Discriminative, Generative and Self-Supervised Approaches for Target-Agnostic Learning
• Dual Prototypical Contrastive Learning for Few-shot Semantic Segmentation
• Contrastive Learning of Image Representations with Cross-Video Cycle-Consistency
• Bridging the Gap Between $f$-GANs and Wasserstein GANs
• Meta-Learning to Improve Pre-Training
• Deep Unsupervised Image Anomaly Detection: An Information Theoretic Framework
• How to Write Summaries with Patterns? Learning towards Abstractive Summarization through Prototype Editing
• Guided Point Contrastive Learning for Semi-supervised Point Cloud Semantic Segmentation
• Global Context Enhanced Social Recommendation with Hierarchical Graph Neural Networks