Key result
Gradient-free sign activation networks require greater adversarial distortion on medical images and ECGs than standard models.
Why the study?
Imperceptible adversarial attacks on medical imaging and ECG data can lead to misdiagnosis and fraud, necessitating models that detect and deter them.
Does a gradient-free trained sign activation neural network ensemble model improve adversarial robustness compared to standard models in medical imaging and ECG classification?
Does a gradient-free trained sign activation neural network ensemble model improve adversarial robustness compared to standard models in medical imaging and ECG classification?
Gradient-free trained sign activation neural networks offer improved robustness against adversarial attacks in medical imaging and ECG classification, potentially enhancing the security of medical AI systems.
Adversarial attacks in medical AI imaging systems can lead to misdiagnosis and insurance fraud as recently highlighted by Finlayson et. al. in Science 2019. They can also be carried out on widely used ECG time-series data as shown in Han et. al. in Nature Medicine 2020. At the heart of adversarial attacks are imperceptible distortions that are visually and statistically undetectable but cause the machine learning model to misclassify data. Recent empirical studies have shown that a gradient-free trained sign activation neural network ensemble model requires a larger distortion than state of the art models. We apply them on medical data in this study as a potential solution to detect and deter adversarial attacks. We show on chest X-ray and histopathology images, and on two ECG datasets that this model requires a greater distortion to be fooled than full-precision, binary, and convolutional neural networks, and random forests. We show that adversaries targeting the gradient-free sign networks are visually distinguishable from the original data and thus likely to be detected by human inspection. Since the sign network distortions are higher we expect an automated method could be developed to detect and deter attacks in advance. Our work here is a significant step towards safe and secure medical machine learning.
No takes yet. Share an insight, caveat, or question.
Yang et al. (2020) studied Adversarial attacks on medical AI systems. Gradient-free trained sign activation neural network ensemble model vs. Full-precision, binary, and convolutional neural networks, and random forests was evaluated on Distortion required to fool the model. Gradient-free trained sign activation neural networks required a greater distortion to be fooled by adversarial attacks on medical images and ECGs compared to standard models.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: