Traditional Adversarial

Traditional adversarial attacks aim to deceive machine learning models by subtly altering input data, causing misclassifications while remaining largely imperceptible to humans. Current research focuses on developing more effective attacks across various data types (images, text, graphs, radio signals) and model architectures (CNNs, GNNs, LLMs, quantum models), often employing techniques like generative models, saliency maps, and semantic manipulation to craft these adversarial examples. This field is crucial for evaluating and improving the robustness of machine learning systems, with implications for security, reliability, and the ethical deployment of AI in diverse applications.

Papers

March 14, 2022

Defending Against Adversarial Attack in ECG Classification with Adversarial Distillation Training
Jiahao Shao, Shijia Geng, Zhaoji Fu, Weilun Xu, Tong Liu, Shenda Hong
Adversarial Attack ECG Classification Traditional Adversarial Adversarial Distillation White Box Adversarial Attack

March 11, 2022

Learning from Attacks: Attacking Variational Autoencoder for Improving Image Classification
Jianzhang Zheng, Fan Yang, Hao Shen, Xuan Tang, Mingsong Chen, Liang Song, Xian Wei
Deep Neural Network Adversarial Attack Adversarial Example Image Classification Variational Autoencoder New Attack Traditional Adversarial

December 16, 2021

TAFIM: Targeted Adversarial Attacks against Facial Image Manipulations
Shivangi Aneja, Lev Markhasin, Matthias Niessner
Adversarial Attack Face Manipulation Traditional Adversarial Image Perturbation Targeted Perturbation

Traditional Adversarial

Papers

Defending Against Adversarial Attack in ECG Classification with Adversarial Distillation Training

Learning from Attacks: Attacking Variational Autoencoder for Improving Image Classification

TAFIM: Targeted Adversarial Attacks against Facial Image Manipulations