Rethinking the Vulnerability of Concept Erasure and a New Method
机构 * School of Mathematics University of Edinburgh(数学学院爱丁堡大学) ; School of Mathematical and Computer Science Heriot-Watt University(数学与计算机科学学院赫瑞斯大学)
高校专区
机构 * School of Mathematics University of Edinburgh(数学学院爱丁堡大学) ; School of Mathematical and Computer Science Heriot-Watt University(数学与计算机科学学院赫瑞斯大学)
机构 * School of Engineering, CHAI Hub University of Edinburgh(爱丁堡大学工程学院、CHAI Hub大学) ; Institute for Analytics and Data Science University of Essex(埃塞克斯大学分析与数据科学研究所) ; Institute for Social and Economic Research University of Essex(埃塞克斯大学社会科学与经济研究所)
Comments Substantially revised version. 18 pages of main content (33 pages in total), 4 main figures (11 in total)