FIMBA: Evaluating the Robustness of AI in Genomics via Feature Importance Adversarial Attacks

TL;DR

This paper evaluates the robustness of AI models in genomics by demonstrating their vulnerability to feature importance adversarial attacks, showing significant performance degradation and proposing spectral analysis for defense insights.

Contribution

It introduces a novel adversarial attack method targeting feature importance in genomics AI models and analyzes adversarial samples to inform countermeasures.

Findings

AI models' accuracy declines under attack

False positives and negatives increase with adversarial samples

Spectral analysis reveals vulnerabilities in model robustness

Abstract

With the steady rise of the use of AI in bio-technical applications and the widespread adoption of genomics sequencing, an increasing amount of AI-based algorithms and tools is entering the research and production stage affecting critical decision-making streams like drug discovery and clinical outcomes. This paper demonstrates the vulnerability of AI models often utilized downstream tasks on recognized public genomics datasets. We undermine model robustness by deploying an attack that focuses on input transformation while mimicking the real data and confusing the model decision-making, ultimately yielding a pronounced deterioration in model performance. Further, we enhance our approach by generating poisoned data using a variational autoencoder-based model. Our empirical findings unequivocally demonstrate a decline in model performance, underscored by diminished accuracy and an upswing in false positives and false negatives. Furthermore, we analyze the resulting adversarial samples via spectral analysis yielding conclusions for countermeasures against such attacks.