Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2601.17093 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918303576883200 |
|---|---|
| author | Sirikova, Olha Chan, Alvin |
| author_facet | Sirikova, Olha Chan, Alvin |
| contents | Comparing neural network representations is essential for understanding and validating models in scientific applications. Existing methods, however, often provide a limited view. We propose the Triangle of Similarity, a framework that combines three complementary perspectives: static representational similarity (CKA/Procrustes), functional similarity (Linear Mode Connectivity or Predictive Similarity), and sparsity similarity (robustness under pruning). Analyzing a range of CNNs, Vision Transformers, and Vision-Language Models using both in-distribution (ImageNetV2) and out-of-distribution (CIFAR-10) testbeds, our initial findings suggest that: (1) architectural family is a primary determinant of representational similarity, forming distinct clusters; (2) CKA self-similarity and task accuracy are strongly correlated during pruning, though accuracy often degrades more sharply; and (3) for some model pairs, pruning appears to regularize representations, exposing a shared computational core. This framework offers a more holistic approach for assessing whether models have converged on similar internal mechanisms, providing a useful tool for model selection and analysis in scientific research. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_17093 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | The Triangle of Similarity: A Multi-Faceted Framework for Comparing Neural Network Representations Sirikova, Olha Chan, Alvin Machine Learning Artificial Intelligence Comparing neural network representations is essential for understanding and validating models in scientific applications. Existing methods, however, often provide a limited view. We propose the Triangle of Similarity, a framework that combines three complementary perspectives: static representational similarity (CKA/Procrustes), functional similarity (Linear Mode Connectivity or Predictive Similarity), and sparsity similarity (robustness under pruning). Analyzing a range of CNNs, Vision Transformers, and Vision-Language Models using both in-distribution (ImageNetV2) and out-of-distribution (CIFAR-10) testbeds, our initial findings suggest that: (1) architectural family is a primary determinant of representational similarity, forming distinct clusters; (2) CKA self-similarity and task accuracy are strongly correlated during pruning, though accuracy often degrades more sharply; and (3) for some model pairs, pruning appears to regularize representations, exposing a shared computational core. This framework offers a more holistic approach for assessing whether models have converged on similar internal mechanisms, providing a useful tool for model selection and analysis in scientific research. |
| title | The Triangle of Similarity: A Multi-Faceted Framework for Comparing Neural Network Representations |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2601.17093 |