Minimal Images in Deep Neural Networks: Fragile Object Recognition in Natural Images
Center for Brains, Minds, and Machines (CBMM)(脑、心智与机器中心(CBMM));Massachusetts Institute of Technology(麻省理工学院);Children’s Hospital, Harvard Medical School(哈佛医学院儿童医院)
The human ability to recognize objects is impaired when the object is not shown in full. "Minimal images" are the smallest regions of an image that remain recognizable for humans. Ullman et al. 2016 show that a slight modification of the location and size of the visible region of the minimal image produces a sharp drop in human recognition accuracy. In this paper, we demonstrate that such drops in accuracy due to changes of the visible region are a common phenomenon between humans and existing state-of-the-art deep neural networks (DNNs), and are much more prominent in DNNs. We found many cases where DNNs classified one region correctly and the other incorrectly, though they only differed by one row or column of pixels, and were often bigger than the average human minimal image size. We show that this phenomenon is independent from previous works that have reported lack of invariance to minor modifications in object location in DNNs. Our results thus reveal a new failure mode of DNNs that also affects humans to a much lesser degree. They expose how fragile DNN recognition ability is for natural images even without adversarial patterns being introduced. Bringing the robustness of DNNs in natural images to the human level remains an open challenge for the community.
University of Notre Dame(圣母大学);Warsaw University of Technology(华沙工业大学);PayEye Poland(波兰佩耶公司);Medical University of Warsaw(华沙医科大学);Indiana University Indianapolis(印第安纳大学印第安纳波利斯分校)
DCA: Dividing and Conquering Amnesia in Incremental Object Detection
Institute of Information Engineering, Chinese Academy of Sciences(中国科学院信息工程研究所);College of Computer Science, Nankai University(南开大学计算机学院);School of Cyber Security, University of Chinese Academy of Sciences(中国科学院大学网络空间安全学院);Harbin Institute of Technology(哈尔滨工业大学);Tsinghua University(清华大学)
A model for interpreting social interactions in local image regions
Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology(麻省理工学院计算机科学与人工智能实验室);Department of Computer Science, Weizmann Institute of Science(魏茨曼科学研究所计算机科学系);Center for Brains Minds and Machines, Massachusetts Institute of Technology(麻省理工学院大脑、心智与机器中心)
Emergence of Painting Ability via Recognition-Driven Evolution
Nanyang Technological University(南洋理工大学);National University of Singapore(新加坡国立大学);RIKEN(理化学研究所);The University of Tokyo(东京大学);Tianjin Academy of Fine Art(天津美术学院);Sakana AI
SphereFace: Deep Hypersphere Embedding for Face Recognition
Georgia Institute of Technology(佐治亚理工学院);Carnegie Mellon University(卡内基梅隆大学);Sun Yat-Sen University(中山大学)
本文研究了开集协议下的深度人脸识别(FR)问题,其中理想的人脸特征在适当选择的度量空间下具有比最小类间距离更小的最大类内距离。然而,很少有现有的算法可以有效地实现这一标准。为此,我们提出了角度softmax(A-Softmax)损失,使卷积神经网络(CNN)能够学习角度判别特征。在几何上,A-Softmax损失可以被视为对超球面流形施加歧视性约束,这本质上匹配面也位于流形上的先验。此外,可以通过参数$m$定量地调整角度裕度的大小。我们进一步推导出具体的$m$来近似理想的特征标准。在Labeled Face in the Wild(LFW)、Youtube Faces(YTF)和MegaFace Challenge上的大量分析和实验表明,A-Softmax损失在FR任务中具有优越性。该守则也已公开提供。
Self-Supervised Pretraining for Fine-Grained Plankton Recognition
LUT University, Computer Vision and Pattern Recognition Laboratory(卢霍斯大学);Finnish Environment Institute(芬兰环境研究所);Brno University of Technology, Faculty of Information Technology(布拉格技术大学)