mlsanityData to 7 Oct 2026

Deconfounded Representation Similarity for Comparison of Neural Networks

arXiv 2202.00095 ↗First posted 31 Jan 2022stat.ML cs.LGcited by 18

Not on the map. Why?

Abstract (arXiv)

Similarity metrics such as representational similarity analysis (RSA) and centered kernel alignment (CKA) have been used to compare layer-wise representations between neural networks. However, these metrics are confounded by the population structure of data items in the input space, leading to spuriously high similarity for even completely random neural networks and inconsistent domain relations in transfer learning. We introduce a simple and generally applicable fix to adjust for the confounder with covariate adjustment regression, which retains the intuitive invariance properties of the original similarity measures. We show that deconfounding the similarity metrics increases the resolution of detecting semantically similar neural networks. Moreover, in real-world applications, deconfounding improves the consistency of representation similarities with domain similarities in transfer learning, and increases correlation with out-of-distribution accuracy.

Key facts. Cited by 18 papers in the map (data through 7 Oct 2026). Builds on 27 of its 54 references that are in the map. First posted on arXiv 31 Jan 2022.

Builds on

27 of 54 references.

References · 27

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  3. 2017A Unified Approach to Interpreting Model Predictionscited by 6,488
  4. 2019EfficientNet: Rethinking Model Scaling for Convolutional Neural Networkscited by 6,415
  5. 2018GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understandingcited by 4,854
  6. 2019DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lightercited by 4,243
  7. 2016SQuAD: 100,000+ Questions for Machine Comprehension of Textcited by 4,115
  8. 2019Unsupervised Cross-lingual Representation Learning at Scalecited by 3,907
  9. 2017Axiomatic Attribution for Deep Networkscited by 3,350
  10. 2019Benchmarking Neural Network Robustness to Common Corruptions and Perturbationscited by 2,857
  11. 2019Similarity of Neural Network Representations Revisitedcited by 1,695
  12. 2014FitNets: Hints for Thin Deep Netscited by 1,619
15 more
  1. 2019CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Comparisoncited by 1,410
  2. 2018Moment Matching for Multi-Source Domain Adaptationcited by 1,327
  3. 2016Understanding intermediate layers using linear classifier probescited by 1,279
  4. 2021Do Vision Transformers See Like Convolutional Neural Networks?cited by 671
  5. 2020What is being transferred in transfer learning?cited by 365
  6. 2018Insights on representational similarity in neural networks with canonical correlationcited by 318
  7. 2020Do Wide and Deep Networks Learn the Same Things? Uncovering How Neural Network Representations Vary with Width and Depthcited by 241
  8. 2018Failing Loudly: An Empirical Study of Methods for Detecting Dataset Shiftcited by 240
  9. 2021Generalized Shape Metrics on Neural Representationscited by 109
  10. 2020Why Do Better Loss Functions Lead to Less Transferable Features?cited by 73
  11. 2020Unraveling Meta-Learning: Understanding Feature Representations for Few-Shot Taskscited by 47
  12. 2021Grounding Representation Similarity with Statistical Testingcited by 28
  13. 2020An analysis of the transfer learning of convolutional neural networks for artistic imagescited by 11
  14. 2020Transferred Discrepancy: Quantifying the Difference Between Representationscited by 9
  15. 2021Disentangling Transfer and Interference in Multi-Domain Learningcited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 3; 2023: 1; 2024: 7; 2025: 3; 2026: 4.

18 papers.

Sort:

  1. 2023Similarity of Neural Network Models: A Survey of Functional and Representational Measurescited by 128
  2. 2022Toward Transparent AI: A Survey on Interpreting the Inner Structures of Deep Neural Networkscited by 126
  3. 2026Revisiting the Platonic Representation Hypothesis: An Aristotelian Viewcited by 33
  4. 2022ModelDiff: A Framework for Comparing Learning Algorithmscited by 30
  5. 2024ReSi: A Comprehensive Benchmark for Representational Similarity Measurescited by 18
  6. 2022GULP: a prediction-based metric between representationscited by 17
  7. 2024Objective drives the consistency of representational similarity across datasetscited by 16
  8. 2025Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environmentscited by 4
  9. 2024Measuring Domain Shifts using Deep Learning Remote Photoplethysmography Model Similaritycited by 2
  10. 2024Refining Remote Photoplethysmography Architectures using CKA and Empirical Methodscited by 1
  11. 2024Reconciling Privacy and Explainability in High-Stakes: A Systematic Inquirycited by 1
  12. 2026Layer-wise Representation Dynamics: An Empirical Investigation Across Embedders and Base LLMscited by 1
6 more
  1. 2026Stimulus symmetries can confound representational similarity analysescited by 1
  2. 2024Has Your Pretrained Model Improved? A Multi-head Posterior Based Approachcited by 0
  3. 2024Revealing Vulnerabilities of Neural Networks in Parameter Learning and Defense Against Explanation-Aware Backdoorscited by 0
  4. 2025ReStNet: A Reusable & Stitchable Network for Dynamic Adaptation on IoT Devicescited by 0
  5. 2025Revisiting the Relation Between Robustness and Universalitycited by 0
  6. 2026Tracing distinguishability through transformer processing with stochastic LayerNormcited by 0