演讲者： Wang Jingdong (Microsoft Research Asia)

时间：2018-05-25 15:00-16:00

地点： 北京大学二教319

Abstract: Eliminating the redundancy in convolution kernels has been attracting increasing interests for designing efficient convolutional neural network architectures with three goals: small model, fast computation, and high accuracy. Existing solutions include low-precision kernels, structured sparse kernels, low-rank kernels, and the product of low-rank kernels. In this talk, I will introduce a novel framework: Interleaved Group Convolution (IGC), which uses the product of structured sparse kernels to compose a dense convolution kernel. It is a drop-in replacement of normal convolution and can be applied to any networks that depend on convolution. I present the complementary condition and the balance condition to guide the design, obtaining a balance between three aspects: model size, computation complexity and classification performance. I will show empirical and theoretic justification of the advantage of the proposed approach over Xception and MobileNet. In addition, IGC raises a rarely-studied matrix decomposition problem: sparse matrix factorization (SMF). I expect more research efforts in SMF from the researchers in the area of matrix analysis.

Bio: Jingdong Wang is a Senior Researcher at the Visual Computing Group, Microsoft Research, Beijing, China. His areas of interest include computer vision, machine learning, and multimedia. He is currently working on CNN architecture design, human understanding, person re-identification, multimedia search, and large-scale indexing. He has served as an area chair (or SPC) in prestigious conferences, including CVPR, ICCV, ECCV, ACMMM, AAAI, IJCAI, and so on. He is an editorial board member for IEEE Transactions on Circuits and Systems for Video Technology, IEEE Transactions on Multimedia, and Tools and Applications. He is a Fellow of the IAPR. He has shipped 10+ technologies to Microsoft products, including XiaoIce Chatbot, Microsoft cognitive service, and Bing search.