analiza aiqbench
This research identifies how latent symmetries in network inputs can distort representational similarity matrices (RSMs), potentially leading to misleading conclusions about neural code geometry. The study demonstrates that standard training methods like stochastic gradient descent may generate drifting codes that obscure the true functional equivalence of representations, complicating comparisons between nonlinear neural networks. These findings are critical for researchers relying on RSMs to interpret and compare deep learning models or biological neural systems.
Stimulus symmetries can confound representational similarity analyses
arXiv:2605.21324v2 Announce Type: replace-cross Abstract: What can representational similarity matrices (RSMs) tell us about a neural code? As the popularity of these summary statistics grows, so too does the need for a more complete characterization of their properties. Here, we show that symmetries in network inputs can confound RSM-based analyses. Stimulus symmetries render many representations functionally equivalent, but these different configurations can lead to different RSMs. These different RSMs reflect qualitatively different representational geometries, ranging from disentangled to maximally-mixed codes. We show that stochastic gradient descent or energetic regularization can generate sparse, drifting codes, leading in turn to drifting RSMs. Moreover, we demonstrate that these phenomena are present in networks trained to encode image data, where the symmetry is latent. Our results illustrate the challenges inherent in comparing nonlinear neural codes, when functionally-equivalent representations are not related by a simple rotation.
Abstract:What can representational similarity matrices (RSMs) tell us about a neural code? As the popularity of these summary statistics grows, so too does the need for a more complete characterization of their properties. Here, we show that symmetries in network inputs can confound RSM-based analyses. Stimulus symmetries render many representations functionally equivalent, but these different configurations can lead to different RSMs. These different RSMs reflect qualitatively different representational geometries, ranging from disentangled to maximally-mixed codes. We show that stochastic gradient descent or energetic regularization can generate sparse, drifting codes, leading in turn to drifting RSMs. Moreover, we demonstrate that these phenomena are present in networks trained to encode image data, where the symmetry is latent. Our results illustrate the challenges inherent in comparing nonlinear neural codes, when functionally-equivalent representations are not related by a simple rotation.
Neurons and Cognition (q-bio.NC); Machine Learning (cs.LG)
arXiv:2605.21324 [q-bio.NC]
(or arXiv:2605.21324v2 [q-bio.NC] for this version)
https://doi.org/10.48550/arXiv.2605.21324 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 232 cuvinte
Subiecte conexe
Scale, Drivers, and Limits of Representational Alignment Across Large Language Models
arXiv · 15 oct. 2026
A Key-Position Confound in no-cot-bench (and implications for interpreting looped transformers)
LessWrong · 2 oct. 2026
Physics-Based Molecular Fingerprints from Spectral Graph Theory Provide Efficient Geometry-Aware Measures of Chemical Similarity
arXiv · 7 aug. 2026
Small LLMs perform no-CoT two-hop reasoning on synthetic facts with real bridges
LessWrong · 17 oct. 2026