Abstract:Deep neural networks have achieved impressive performance on a variety of tasks, but their brittleness to distributional shifts remains a significant barrier to real-world deployment. In this paper, we propose a framework to analyse and quantify the distributional robustness of neural networks by studying the interactions between layer weights and activations. We model these interactions using Bernoulli distributions, using the separation between classes as a diagnostic proxy for robustness. We demonstrate the usefulness of this framework through models trained on CIFAR-10 and ImageNet. We show that our proposed metrics can distinguish between networks that have memorised their training data and those that have not. We also perform analogous experiments in the activation space and find that the same properties do not hold up. Additionally, we investigate the behaviour of our metrics under various distribution shifts and show that these shifts reduce separation under our path-based diagnostics. Our results suggest that this framework provides useful model-level diagnostics of representation structure and robustness.
| Comments: | 9 pages, 6 figures, 3 tables |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2605.21313 [cs.LG] |
| (or arXiv:2605.21313v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2605.21313 arXiv-issued DOI via DataCite (pending registration) |
From: Divij Khaitan [view email]
[v1]
Wed, 20 May 2026 15:42:56 UTC (6,080 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。