• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>IEEE International System-on-Chip Conference >T1B: The Good, The Bad, and The Potential, A Tutorial for the Adversarial Attacks
【24h】

T1B: The Good, The Bad, and The Potential, A Tutorial for the Adversarial Attacks

机译：T1B：良好，坏，潜力，潜在，对抗攻击的教程

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

With Deep Neural Network (DNN) as the most representative technology, the machine learning has the been widely applied in to daily applications, However, with the fast -growth of machine learning technology and DNN equipped applications, a new DNN attack method - adversarial attack has demonstrated considerable negative impact to the machine learning development. The adversarial attack is derived from the DNN model structure to manipulate the classification result by crafting dedicated perturbations on the original images, that even human vision can't perceive. Recently many research works are focusing on the adversarial attack generation and defense.
机译：随着深度神经网络（DNN）作为最具代表性的技术，机器学习已被广​​泛应用于日常应用，然而，随着机器学习技术的快速生长和DNN装备的应用，一种新的DNN攻击方法 - 对抗攻击对机器学习开发表现出相当大的负面影响。对抗攻击来自DNN模型结构来通过制定在原始图像上的专用扰动来操纵分类结果，即使人类的视力也无法感知。最近许多研究作品专注于对抗生成和防御。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号