Friendly Adversarial Training: Attacks Which Do Not Kill Training Make Adversarial Learning Stronger
Jul 12, 2020
Sprecher:innen
Jingfeng Zhang
Sprecher:in · 0 Follower:innen
Masashi Sugiyama
Sprecher:in · 2 Follower:innen
Mohan Kankanhalli
Sprecher:in · 0 Follower:innen
Lichen Cui
Sprecher:in · 0 Follower:innen
Über
Adversarial training based on the minimax formulation is necessary for obtaining adversarial robustness of trained models. However, it is conservative or even pessimistic so that it sometimes hurts the natural generalization. In this paper, we raise a fundamental question—do we have to trade off natural generalization for adversarial robustness? We argue that adversarial training is to employ confident adversarial data for updating the current model. We propose a novel approach of friendly adversarial training (FAT): rather than employing most adversarial data maximizing the loss, we search for least adversarial (i.e., friendly adversarial) data minimizing the loss, among the adversarial data that are confidently misclassified. Our novel formulation is easy to implement by just stopping the most-adversarial data searching algorithms such as PGD (projected gradient descent) early, which we call early-stopped PGD. Theoretically, FAT is justified by an upper bound of the adversarial risk. Empirically, early-stopped PGD allows us to answer the earlier question negatively—adversarial robustness can indeed be achieved without compromising the natural generalization.Adversarial training based on the minimax formulation is necessary for obtaining adversarial robustness of trained models. However, it is conservative or even pessimistic so that it sometimes hurts the natural generalization. In this paper, we raise a fundamental question—do we have to trade off natural generalization for adversarial robustness? We argue that adversarial training is to employ confident adversarial data for updating the current model. We propose a novel approach of friendly adver…
Kategorien
KI und Datenwissenschaft
Kategorie · 10,8k Präsentationen
Über ICML 2020
The International Conference on Machine Learning (ICML) is the premier gathering of professionals dedicated to the advancement of the branch of artificial intelligence known as machine learning. ICML is globally renowned for presenting and publishing cutting-edge research on all aspects of machine learning used in closely related areas like artificial intelligence, statistics and data science, as well as important application areas such as machine vision, computational biology, speech recognition, and robotics. ICML is one of the fastest growing artificial intelligence conferences in the world. Participants at ICML span a wide range of backgrounds, from academic and industrial researchers, to entrepreneurs and engineers, to graduate students and postdocs.
Gefällt euch das Format? Vertraut auf SlidesLive, um euer nächstes Event festzuhalten!
Professionelle Aufzeichnung und Livestreaming – weltweit.
Freigeben
Empfohlene Videos
Präsentationen, deren Thema, Kategorie oder Sprecher:in ähnlich sind
On the Consistency of Top-k Surrogate Losses
Später ansehen
Forest Yang, …
Do RNN and LSTM have Long Memory?
Später ansehen
Jingyu Zhao, …
Poster #30
Später ansehen
Task-Oriented Active Perception and Planning in Environments with Partially Known Semantics
Später ansehen
Curvature-guided Pruning of High-performance Neural Networks Using Ricci Flow
Später ansehen
Samuel Glass, …