Tianyu Ding

Tianyu Ding is a principal researcher in Microsoft’s Applied Sciences Group (ASG), working on computer vision problems with deep-learning tools.  Before joining Microsoft, he received his Ph.D. degree from Johns Hopkins University in 2021, where his research focused on subspace learning for data arising from a union of subspaces of high dimension.  He has a broad interest in developing cutting-edge computer vision techniques.

During his research internships at Microsoft (2020 and 2021), he worked on designing efficient deep neural networks based on model compression for video frame interpolation and novel view synthesis.

• OTOV2: Automatic, Generic, User-Friendly

  The Eleventh International Conference on Learning Representations (ICLR), 2023 June, 2023 Pages 1-25

  By Tianyi Chen, Luming Liang, Tianyu Ding, Ilya Zharkov, Zhihui Zhu
• RSTT: Real-time Spatial Temporal Transformer for Space-Time Video Super-Resolution

  IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022 June, 2022 Pages 17441-17451

  By Zhicheng Geng, Luming Liang, Tianyu Ding, Ilya Zharkov
• Only Train Once: A One-Shot Neural Network Training And Pruning Framework

  35th Conference on Neural Information Processing Systems (NeurIPS 2021) July, 2021

  By Tianyi Chen, Bo Ji, Tianyu Ding, Biyi Fang, Guanyi Wang, Zhihui Zhu, Luming Liang, Yixin Shi, Sheng Yi, Xiao Tu
• CDFI: Compression-Driven Network Design for Frame Interpolation

  Computer Vision and Pattern Recognition (CVPR), 2021 June, 2021 Pages 8001-8011

  By Tianyu Ding, Luming Liang, Zhihui Zhu, Ilya Zharkov