Representations and generalization in artificial and brain neural networks

Q Li, B Sorscher, H Sompolinsky - Proceedings of the National Academy of …, 2024 - pnas.org
Humans and animals excel at generalizing from limited data, a capability yet to be fully
replicated in artificial intelligence. This perspective investigates generalization in biological …

Modeling the influence of data structure on learning in neural networks: The hidden manifold model

S Goldt, M Mézard, F Krzakala, L Zdeborová - Physical Review X, 2020 - APS
Understanding the reasons for the success of deep neural networks trained using stochastic
gradient-based methods is a key open problem for the nascent theory of deep learning. The …

Separability and geometry of object manifolds in deep neural networks

U Cohen, SY Chung, DD Lee, H Sompolinsky - Nature communications, 2020 - nature.com
Stimuli are represented in the brain by the collective population responses of sensory
neurons, and an object presented under varying conditions gives rise to a collection of …

The gaussian equivalence of generative models for learning with shallow neural networks

S Goldt, B Loureiro, G Reeves… - Mathematical and …, 2022 - proceedings.mlr.press
Understanding the impact of data structure on the computational tractability of learning is a
key challenge for the theory of neural networks. Many theoretical works do not explicitly …

Classification and geometry of general perceptual manifolds

SY Chung, DD Lee, H Sompolinsky - Physical Review X, 2018 - APS
Perceptual manifolds arise when a neural population responds to an ensemble of sensory
signals associated with different physical features (eg, orientation, pose, scale, location, and …

A unified theory of early visual representations from retina to cortex through anatomically constrained deep CNNs

J Lindsey, SA Ocko, S Ganguli, S Deny - arXiv preprint arXiv:1901.00945, 2019 - arxiv.org
The visual system is hierarchically organized to process visual information in successive
stages. Neural representations vary drastically across the first stages of visual processing: at …

Linear classification of neural manifolds with correlated variability

AJ Wakhloo, TJ Sussman, SY Chung - Physical Review Letters, 2023 - APS
Understanding how the statistical and geometric properties of neural activity relate to
performance is a key problem in theoretical neuroscience and deep learning. Here, we …

Beyond the storage capacity: data-driven satisfiability transition

P Rotondo, M Pastore, M Gherardi - Physical Review Letters, 2020 - APS
Data structure has a dramatic impact on the properties of neural networks, yet its
significance in the established theoretical frameworks is poorly understood. Here we …

Counting the learnable functions of geometrically structured data

P Rotondo, MC Lagomarsino, M Gherardi - Physical Review Research, 2020 - APS
Cover's function counting theorem is a milestone in the theory of artificial neural networks. It
provides an answer to the fundamental question of determining how many binary …

Attributed graph force learning

K Sun, F Xia, J Liu, B Xu, V Saikrishna… - … on Neural Networks …, 2022 - ieeexplore.ieee.org
In numerous network analysis tasks, feature representation plays an imperative role. Due to
the intrinsic nature of networks being discrete, enormous challenges are imposed on their …