# Identifying Audio Adversarial Examples via Anomalous Pattern Detection

arXiv: https://arxiv.org/abs/2002.05463 · HTML: https://mlsanity.com/paper/2002.05463/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Feb 2020.

## Built on by (most cited first, 9 of 9)

- [Weakly Supervised Detection of Hallucinations in LLM Activations](https://mlsanity.com/paper/2312.02798/) (2023)
- [Towards creativity characterization of generative models via group-based subset scanning](https://mlsanity.com/paper/2104.00479/) (2021)
- [Pattern Detection in the Activation Space for Identifying Synthesized Content](https://mlsanity.com/paper/2105.12479/) (2021)
- [SoK: A Modularized Approach to Study the Security of Automatic Speech Recognition Systems](https://mlsanity.com/paper/2103.10651/) (2021)
- [Localizing Persona Representations in LLMs](https://mlsanity.com/paper/2505.24539/) (2025)
- [Leveraging Domain Features for Detecting Adversarial Attacks Against Deep Speech Recognition in Noise](https://mlsanity.com/paper/2211.01621/) (2022)
- [Efficient Representation of the Activation Space in Deep Neural Networks](https://mlsanity.com/paper/2312.08143/) (2023)
- [SoK: A Study of the Security on Voice Processing Systems](https://mlsanity.com/paper/2112.13144/) (2021)
- [Towards Creativity Characterization of Generative Models via Group-based Subset Scanning](https://mlsanity.com/paper/2203.00523/) (2022)

## Abstract (arXiv)

Audio processing models based on deep neural networks are susceptible to adversarial attacks even when the adversarial audio waveform is 99.9% similar to a benign sample. Given the wide application of DNN-based audio recognition systems, detecting the presence of adversarial examples is of high practical relevance. By applying anomalous pattern detection techniques in the activation space of these models, we show that 2 of the recent and current state-of-the-art adversarial attacks on audio processing systems systematically lead to higher-than-expected activation at some subset of nodes and we can detect these with up to an AUC of 0.98 with no degradation in performance on benign samples.