On the Computational Entanglement of Distant Features in Adversarial Machine Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lai, YenLung, Dong, Xingbo, Jin, Zhe
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910623858688000
author Lai, YenLung
Dong, Xingbo
Jin, Zhe
author_facet Lai, YenLung
Dong, Xingbo
Jin, Zhe
contents In this research, we introduce the concept of "computational entanglement," a phenomenon observed in overparameterized feedforward linear networks that enables the network to achieve zero loss by fitting random noise, even on previously unseen test samples. Analyzing this behavior through spacetime diagrams reveals its connection to length contraction, where both training and test samples converge toward a shared normalized point within a flat Riemannian manifold. Moreover, we present a novel application of computational entanglement in transforming a worst-case adversarial examples-inputs that are highly non-robust and uninterpretable to human observers-into outputs that are both recognizable and robust. This provides new insights into the behavior of non-robust features in adversarial example generation, underscoring the critical role of computational entanglement in enhancing model robustness and advancing our understanding of neural networks in adversarial contexts.
format Preprint
id arxiv_https___arxiv_org_abs_2309_15669
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle On the Computational Entanglement of Distant Features in Adversarial Machine Learning
Lai, YenLung
Dong, Xingbo
Jin, Zhe
Machine Learning
Information Theory
Computational Physics
In this research, we introduce the concept of "computational entanglement," a phenomenon observed in overparameterized feedforward linear networks that enables the network to achieve zero loss by fitting random noise, even on previously unseen test samples. Analyzing this behavior through spacetime diagrams reveals its connection to length contraction, where both training and test samples converge toward a shared normalized point within a flat Riemannian manifold. Moreover, we present a novel application of computational entanglement in transforming a worst-case adversarial examples-inputs that are highly non-robust and uninterpretable to human observers-into outputs that are both recognizable and robust. This provides new insights into the behavior of non-robust features in adversarial example generation, underscoring the critical role of computational entanglement in enhancing model robustness and advancing our understanding of neural networks in adversarial contexts.
title On the Computational Entanglement of Distant Features in Adversarial Machine Learning
topic Machine Learning
Information Theory
Computational Physics
url https://arxiv.org/abs/2309.15669