# An Information-Theoretic Explanation for the Adversarial Fragility of AI Classifiers

arXiv: https://arxiv.org/abs/1901.09413 · HTML: https://mlsanity.com/paper/1901.09413/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 27 Jan 2019.

## Built on by (most cited first, 5 of 5)

- [Adversarial Examples Are Not Bugs, They Are Superposition](https://mlsanity.com/paper/2508.17456/) (2025)
- [Mutual Information Learned Classifiers: an Information-theoretic Viewpoint of Training Deep Learning Classification Systems](https://mlsanity.com/paper/2209.10058/) (2022)
- [Feature compression is the root cause of adversarial fragility in neural network classifiers](https://mlsanity.com/paper/2406.16200/) (2024)
- [Mutual Information Learned Classifiers: an Information-theoretic Viewpoint of Training Deep Learning Classification Systems](https://mlsanity.com/paper/2210.01000/) (2022)
- [The Endless Tuning. An Artificial Intelligence Design To Avoid Human Replacement and Trace Back Responsibilities](https://mlsanity.com/paper/2507.14909/) (2025)

## Abstract (arXiv)

We present a simple hypothesis about a compression property of artificial intelligence (AI) classifiers and present theoretical arguments to show that this hypothesis successfully accounts for the observed fragility of AI classifiers to small adversarial perturbations. We also propose a new method for detecting when small input perturbations cause classifier errors, and show theoretical guarantees for the performance of this detection method. We present experimental results with a voice recognition system to demonstrate this method. The ideas in this paper are motivated by a simple analogy between AI classifiers and the standard Shannon model of a communication system.