Data to 7 Oct 2026

Learning Diverse and Discriminative Representations via the Principle of Maximal Coding Rate Reduction

arXiv 2006.08558 ↗First posted 15 Jun 2020cs.LG cs.CV cs.ITcited by 141

Not on the map. Why?

Abstract (arXiv)

To learn intrinsic low-dimensional structures from high-dimensional data that most discriminate between classes, we propose the principle of Maximal Coding Rate Reduction (MCR2), an information-theoretic measure that maximizes the coding rate difference between the whole dataset and the sum of each individual class. We clarify its relationships with most existing frameworks such as cross-entropy, information bottleneck, information gain, contractive and contrastive learning, and provide theoretical guarantees for learning diverse and discriminative features. The coding rate can be accurately computed from finite samples of degenerate subspace-like distributions and can learn intrinsic representations in supervised, self-supervised, and unsupervised settings in a unified manner. Empirically, the representations learned using this principle alone are significantly more robust to label corruptions in classification than those using cross-entropy, and can lead to state-of-the-art results in clustering mixed data from self-learned invariant features.

Key facts. Cited by 141 papers in the map (data through 7 Oct 2026). First posted on arXiv 15 Jun 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 14; 2022: 28; 2023: 37; 2024: 25; 2025: 16; 2026: 21.

141 papers.

Sort:

By area:

2021A Geometric Analysis of Neural Collapse with Unconstrained Featurescited by 226
2022Fake it till you make it: Learning transferable representations from synthetic ImageNet clonescited by 171
2021Exploring Deep Neural Networks via Layer-Peeled Model: Minority Collapse in Imbalanced Trainingcited by 166
2022Neural Collapse: A Review on Modelling Principles and Generalizationcited by 95
2022Robust Training under Label Noise by Over-parameterizationcited by 93
2022Are All Losses Created Equal: A Neural Collapse Perspectivecited by 78
2023White-Box Transformers via Sparse Rate Reductioncited by 77
2023D2 Pruning: Message Passing for Balancing Diversity and Difficulty in Data Pruningcited by 68
2022On the Principles of Parsimony and Self-Consistency for the Emergence of Intelligencecited by 50
2022Neural Collapse with Normalized Features: A Geometric Analysis over the Riemannian Manifoldcited by 50
2024Usable XAI: 10 Strategies Towards Exploiting Explainability in the LLM Eracited by 45
2021Frustratingly Easy Transferability Estimationcited by 44

38 more

2022Neural Manifold Clustering and Embeddingcited by 44
2023Generalized Neural Collapse for a Large Number of Classescited by 39
2022Self-Supervised Learning via Maximum Entropy Codingcited by 33
2023Learning Efficient Coding of Natural Images with Maximum Manifold Capacity Representationscited by 31
2025Auto-Rubric: Learning From Implicit Weights to Explicit Rubrics for Reward Modelingcited by 29
2024DGR-MIL: Exploring Diverse Global Representation in Multiple Instance Learning for Whole Slide Image Classificationcited by 25
2021Learning a Self-Expressive Network for Subspace Clusteringcited by 24
2021Clustering by Maximizing Mutual Information Across Viewscited by 23
2022No Reason for No Supervision: Improved Generalization in Supervised Modelscited by 23
2023Emergence of Segmentation with Minimalistic White-Box Transformerscited by 23
2023Image Clustering via the Principle of Rate Reduction in the Age of Pretrained Modelscited by 22
2023Understanding Augmentation-based Self-Supervised Representation Learning via RKHS Approximation and Regressioncited by 20
2025Simplifying DINO via Coding Rate Regularizationcited by 20
2022Localized Adversarial Domain Generalizationcited by 19
2023Interpretable Deep Clustering for Tabular Datacited by 19
2023Is forgetting less a good inductive bias for forward transfer?cited by 18
2023EMP-SSL: Towards Self-Supervised Learning in One Training Epochcited by 18
2023Class Incremental Learning with Self-Supervised Pre-Training and Prototype Learningcited by 18
2021Closed-Loop Data Transcription to an LDR via Minimaxing Rate Reductioncited by 17
2022Incremental Learning of Structured Memory via Closed-Loop Transcriptioncited by 17
2024Latent Intrinsics Emerge from Training to Relightcited by 17
2021MAE-DET: Revisiting Maximum Entropy Principle in Zero-Shot NAS for Efficient Object Detectioncited by 16
2023Information Flow in Self-Supervised Learningcited by 16
2022Learning Fair Representations via Rate-Distortion Maximizationcited by 15
2021Understanding Square Loss in Training Overparametrized Neural Network Classifierscited by 14
2023DeepMAD: Mathematical Architecture Design for Deep Convolutional Neural Networkcited by 14
2023Neural Collapse in Multi-label Learning with Pick-all-label Losscited by 14
2024Guiding Neural Collapse: Optimising Towards the Nearest Simplex Equiangular Tight Framecited by 14
2025Data Pruning by Information Maximizationcited by 14
2023KECOR: Kernel Coding Rate Maximization for Active 3D Object Detectioncited by 13
2024A Global Geometric Analysis of Maximal Coding Rate Reductioncited by 13
2023Unsupervised Manifold Linearizing and Clusteringcited by 12
2023A New Benchmark: On the Utility of Synthetic Data with Blender for Bare Supervised Learning and Downstream Domain Adaptationcited by 11
2023Representation Learning via Manifold Flattening and Reconstructioncited by 11
2023ERASE: Error-Resilient Representation Learning on Graphs for Label Noise Tolerancecited by 11
2021Understanding Neural Networks with Logarithm Determinant Entropy Estimatorcited by 10
2023Mitigating Label Noise through Data Ambiguationcited by 10
2023Robust Concept Erasure via Kernelized Rate-Distortion Maximizationcited by 10

50 most cited of 141 All 141 (JSON)