A Neuromorphic Sparse Coding Defense to Adversarial Images

A Neuromorphic Sparse Coding Defense to Adversarial Images
复制标题

针对对抗性图像的神经形态稀疏编码防御

DOI:
10.1145/3354265.3354277
复制
发表时间:
2019
期刊:
ICONS '19: Proceedings of the International Conference on Neuromorphic Systems
影响因子:
--
通讯作者:
Kenyon, Garrett T.
Kenyon, Garrett T.
中科院分区:
--
文献类型:
--
作者:
Kim, Edward;Yarnall, Jessica;Shah, Priya;Kenyon, Garrett T.

文献摘要

参考文献

被引文献

相似文献

对抗性图像是一类被非常特定的噪声稍微改变的图像,以改变深度学习神经网络对图像进行分类的方式。在许多情况下,这种特殊的噪声对人类视觉系统来说是不可感知的,因此对机器学习和人工智能社区来说是一个值得关注的漏洞。减轻这种类型攻击的研究采取了多种形式,其中之一是在使用深度神经网络对图像进行分类之前对图像进行过滤或后处理。平滑、滤波和压缩等技术已经取得了不同程度的成功。在我们的工作中,我们探索了使用神经形态软件和硬件方法来防止对抗性图像攻击。管理我们的神经形态方法的算法是基于稀疏编码。我们的稀疏编码方法是解决使用一个动态系统的方程模型生物低层次的视觉。我们的定量和定性的结果表明,稀疏编码重建是显着不变的稀疏性和重建误差的变化方面的分类精度。此外,我们的方法是能够保持低重建错误,而不牺牲分类性能。
Adversarial images are a class of images that have been slightly altered by very specific noise to change the way a deep learning neural network classifies the image. In many cases, this particular noise is imperceptible to the human vision system and thus presents a vulnerability of significant concern to the machine learning and artificial intelligence community. Research towards mitigating this type of attack has taken many forms, one of which is to filter or post process the image before classifying the image with a deep neural network. Techniques such as smoothing, filtering, and compression have been used with varying levels of success.In our work, we explored the use of a neuromorphic software and hardware approach as a protection against adversarial image attack. The algorithm governing our neuromorphic approach is based upon sparse coding. Our sparse coding approach is solved using a dynamic system of equations that models biological low level vision. Our quantitative and qualitative results show that a sparse coding reconstruction is remarkably invariant to changes in sparsity and reconstruction error with respect to classification accuracy. Furthermore, our approach is able to maintain low reconstruction errors without sacrificing classification performance.
DOI: 10.1109/isit.2018.8437638
发表时间: 2018-01
期刊: 2018 IEEE International Symposium on Information Theory (ISIT)
影响因子: --
作者:
Zhinus Marzi;S. Gopalakrishnan;Upamanyu Madhow;Ramtin Pedarsani
通讯作者: Zhinus Marzi;S. Gopalakrishnan;Upamanyu Madhow;Ramtin Pedarsani
不变多模态 Halle Berry 神经元的深度稀疏编码
DOI: 10.1109/cvpr.2018.00122
发表时间: 2017
期刊: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition
影响因子: --
作者:
Edward Kim;Darryl Hannan;Garrett T. Kenyon
通讯作者: Garrett T. Kenyon
人工智能白皮书2017
DOI: --
发表时间: 2017
期刊:
影响因子: --
作者:
Ninomiya;Y.;Zhao;W. and Saga;Y.;独立行政法人情報処理推進機構 AI白書編集委員会
通讯作者: 独立行政法人情報処理推進機構 AI白書編集委員会
DOI: 10.48550/arxiv.1608.00853
发表时间: 2016
期刊: --
影响因子: --
作者:
Dziugaite G
通讯作者: Dziugaite G
DOI: 10.1109/icip.2007.4379981
发表时间: 2007
期刊: 2007 IEEE International Conference on Image Processing
影响因子: --
作者:
C. Rozell;Don H. Johnson;Richard Baraniuk;B. Olshausen
通讯作者: B. Olshausen