ADVERSARIAL MACHINE LEARNING: A UNIFIED TWO-DIMENSIONAL LIFECYCLE-PARADIGM FRAMEWORK: FROM FOUNDATION TO FRONTIERS

  • 予約

ADVERSARIAL MACHINE LEARNING: A UNIFIED TWO-DIMENSIONAL LIFECYCLE-PARADIGM FRAMEWORK: FROM FOUNDATION TO FRONTIERS

  • ただいまウェブストアではご注文を受け付けておりません。 ⇒古書を探す
  • 製本 Hardcover:ハードカバー版/ページ数 340 p.
  • 言語 ENG
  • 商品コード 9789819837809

Full Description

This book introduces a unified two-dimensional framework approach to adversarial machine learning—a central pillar to defending deep learning and large language model (LLM) systems against backdoor attacks, adversarial examples, and other deliberate manipulation.Spanning the machine learning lifecycle (pre-training to inference) and three major attack paradigms (backdoor, weight, and adversarial example), the framework introduced is built on the core principles of stealthiness, benign consistency, and adversarial inconsistency. Written for graduate students, researchers, and practitioners, the book also covers constructive applications (copyright and privacy protection, data-abuse mitigation, LLM alignment) and the challenges of generative AI and autonomous agents.

最近チェックした商品