Robustness and Visual Explanation for Black Box Image, Video, and ECG Signal Classification with Reinforcement Learning
This provides a comprehensive tool for improving robustness and transparency in black-box classification systems across multiple domains, though it appears to be an incremental advancement in adversarial attack methods.
The researchers tackled the problem of creating adversarial attacks across different data types (ECG signals, images, videos) by developing a Reinforcement Learning framework that identifies sensitive regions and induces misclassifications with minimal distortions. Their method outperformed state-of-the-art approaches in all three applications, producing superior localization masks that enhance interpretability for models.
We present a generic Reinforcement Learning (RL) framework optimized for crafting adversarial attacks on different model types spanning from ECG signal analysis (1D), image classification (2D), and video classification (3D). The framework focuses on identifying sensitive regions and inducing misclassifications with minimal distortions and various distortion types. The novel RL method outperforms state-of-the-art methods for all three applications, proving its efficiency. Our RL approach produces superior localization masks, enhancing interpretability for image classification and ECG analysis models. For applications such as ECG analysis, our platform highlights critical ECG segments for clinicians while ensuring resilience against prevalent distortions. This comprehensive tool aims to bolster both resilience with adversarial training and transparency across varied applications and data types.