Reinforcement Learning Based Black-Box Adversarial Attack for Robustness Improvement

Sarkar, Soumyendu; Babu, Ashwin Ramesh; Mousavi, Sajad; Ghorbanpour, Sahand; Gundecha, Vineet; Gutierrez, Ricardo Luna; Guillen, Antonio; Naug, Avisek

doi:10.1109/case56687.2023.10260607

Cited by 2 publications

(1 citation statement)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Also, the effectiveness of localization is evaluated with metrics such as dice coefficient and IOU and compared with the popular gradient and non-gradient-based approaches (Selvaraju et al 2017;Ramaswamy et al 2020;Sarkar et al 2023a), with the proposed method showing superiority over the other approaches. Also, retraining the model with adversarial samples significantly improved robustness when evaluated on benchmark datasets (Sarkar et al 2023c).…”

Section: Resultsmentioning

confidence: 99%

Robustness and Visual Explanation for Black Box Image, Video, and ECG Signal Classification with Reinforcement Learning

Sarkar,

Ramesh Babu,

Mousavi

et al. 2024

AAAI

View full text Add to dashboard Cite

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.

show abstract

Section: Resultsmentioning

confidence: 99%