Geometric origin of adversarial vulnerability in deep learning

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Ren, Yixiong, Du, Wenkang, Zhou, Jianhui, Huang, Haiping
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918135574036480
author Ren, Yixiong
Du, Wenkang
Zhou, Jianhui
Huang, Haiping
author_facet Ren, Yixiong
Du, Wenkang
Zhou, Jianhui
Huang, Haiping
contents How to balance training accuracy and adversarial robustness has become a challenge since the birth of deep learning. Here, we introduce a geometry-aware deep learning framework that leverages layer-wise local training to sculpt the internal representations of deep neural networks. This framework promotes intra-class compactness and inter-class separation in feature space, leading to manifold smoothness and adversarial robustness against white or black box attacks. The performance can be explained by an energy model with Hebbian coupling between elements of the hidden representation. Our results thus shed light on the physics of learning in the direction of alignment between biological and artificial intelligence systems. Using the current framework, the deep network can assimilate new information into existing knowledge structures while reducing representation interference.
format Preprint
id arxiv_https___arxiv_org_abs_2509_01235
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Geometric origin of adversarial vulnerability in deep learning
Ren, Yixiong
Du, Wenkang
Zhou, Jianhui
Huang, Haiping
Machine Learning
Statistical Mechanics
Neurons and Cognition
How to balance training accuracy and adversarial robustness has become a challenge since the birth of deep learning. Here, we introduce a geometry-aware deep learning framework that leverages layer-wise local training to sculpt the internal representations of deep neural networks. This framework promotes intra-class compactness and inter-class separation in feature space, leading to manifold smoothness and adversarial robustness against white or black box attacks. The performance can be explained by an energy model with Hebbian coupling between elements of the hidden representation. Our results thus shed light on the physics of learning in the direction of alignment between biological and artificial intelligence systems. Using the current framework, the deep network can assimilate new information into existing knowledge structures while reducing representation interference.
title Geometric origin of adversarial vulnerability in deep learning
topic Machine Learning
Statistical Mechanics
Neurons and Cognition
url https://arxiv.org/abs/2509.01235