mlsanityData to 14 Oct 2026

Representation Alignment in Neural Networks

arXiv 2112.07806 ↗First posted 15 Dec 2021cs.LG cs.AIcited by 6

Not on the map. Why?

Abstract (arXiv)

It is now a standard for neural network representations to be trained on large, publicly available datasets, and used for new problems. The reasons for why neural network representations have been so successful for transfer, however, are still not fully understood. In this paper we show that, after training, neural network representations align their top singular vectors to the targets. We investigate this representation alignment phenomenon in a variety of neural network architectures and find that (a) alignment emerges across a variety of different architectures and optimizers, with more alignment arising from depth (b) alignment increases for layers closer to the output and (c) existing high-performance deep CNNs exhibit high levels of alignment. We then highlight why alignment between the top singular vectors and the targets can speed up learning and show in a classic synthetic transfer problem that representation alignment correlates with positive and negative transfer to similar and dissimilar tasks.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Builds on 39 of its 58 references that are in the map. First posted on arXiv 15 Dec 2021.

Builds on

39 of 58 references.

References · 39

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  3. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  4. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,498
  5. 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 7,005
  6. 2013Distributed Representations of Words and Phrases and their Compositionalitycited by 3,474
  7. 2018Neural Tangent Kernel: Convergence and Generalization in Neural Networkscited by 2,621
  8. 2013Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Mapscited by 2,371
  9. 2013Visualizing and Understanding Convolutional Networkscited by 2,290
  10. 2016Understanding deep learning requires rethinking generalizationcited by 1,674
  11. 2016Understanding intermediate layers using linear classifier probescited by 1,353
  12. 2014How transferable are features in deep neural networks?cited by 1,225
27 more
  1. 2021Tokens-to-Token ViT: Training Vision Transformers from Scratch on ImageNetcited by 1,101
  2. 2018Reconciling modern machine learning practice and the bias-variance trade-offcited by 927
  3. 2020Prevalence of Neural Collapse during the terminal phase of deep learning trainingcited by 719
  4. 2019Deep Double Descent: Where Bigger Models and More Data Hurtcited by 672
  5. 2019Fine-Grained Analysis of Optimization and Generalization for Overparameterized Two-Layer Neural Networkscited by 566
  6. 2018Do Better ImageNet Models Transfer Better?cited by 565
  7. 2015Train faster, generalize better: Stability of stochastic gradient descentcited by 552
  8. 2019Benign Overfitting in Linear Regressioncited by 549
  9. 2019Fantastic Generalization Measures and Where to Find Themcited by 513
  10. 2017Opening the Black Box of Deep Neural Networks via Informationcited by 503
  11. 2020What is being transferred in transfer learning?cited by 368
  12. 2018Insights on representational similarity in neural networks with canonical correlationcited by 324
  13. 2018Characterizing and Avoiding Negative Transfercited by 229
  14. 2018Deep Learning: A Critical Appraisalcited by 217
  15. 2016Fine-grained Analysis of Sentence Embeddings Using Auxiliary Prediction Taskscited by 216
  16. 2014Object Detectors Emerge in Deep Scene CNNscited by 179
  17. 2017What do Neural Machine Translation Models Learn about Morphology?cited by 166
  18. 2020Spectral Bias and Task-Model Alignment Explain Generalization in Kernel Regression and Infinitely Wide Neural Networkscited by 166
  19. 2020What shapes feature representations? Exploring datasets, architectures, and trainingcited by 126
  20. 2018What Is One Grain of Sand in the Desert? Analyzing Individual Neurons in Deep NLP Modelscited by 124
  21. 2018An analytic theory of generalization dynamics and transfer learning in deep linear networkscited by 85
  22. 2016Graying the black box: Understanding DQNscited by 68
  23. 2019Generalization Guarantees for Neural Networks via Harnessing the Low-rank Structure of the Jacobiancited by 67
  24. 2020What Do Neural Networks Learn When Trained With Random Labels?cited by 59
  25. 2021What can linearized neural networks actually say about generalization?cited by 49
  26. 2019Neural Spectrum Alignment: Empirical Studycited by 6
  27. 2020Implicit Regularization via Neural Feature Alignmentcited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2022: 1; 2023: 0; 2024: 2; 2025: 2; 2026: 1.

6 papers.

  1. 2024Training the Untrainable: Introducing Inductive Bias via Representational Alignmentcited by 4
  2. 2024Enabling Multi-Robot Collaboration from Single-Human Guidancecited by 3
  3. 2025Spectral Alignment as Predictor of Loss Explosion in Neural Network Trainingcited by 2
  4. 2022Label Alignment Regularization for Distribution Shiftcited by 1
  5. 2025Model alignment using inter-modal bridgescited by 1
  6. 2026Value Bonuses using Ensemble Errors for Exploration in Reinforcement Learningcited by 1