革新知能統合研究センター

投稿日：5/21（改訂6/7、7/21）

機械学習のトップカンファレンスである International Conference on Machine Learning (ICML) 2021において、AIPセンターから26本の論文が採択されました。

[Website] https://icml.cc/Conferences/2021

[Accepted Papers] https://icml.cc/Conferences/2021/AcceptedPapersInitial

採択数国内1位
AIPセンターは、ICML2021の論文採択数26本*（全採択数1184中2.2%)で組織別にみると国内第1位*、著者別にみると杉山将センター長が採択論文数14本*で全体で第1位*でした。
参考：ICML2021では、5513本の論文投稿があり、採択率21.4％でした。
（*数値はすべて理研AIP調べ2021年7月21日現在）

Long talk (2 papers)

• Confidence Scores Make Instance-dependent Label-noise Learning Possible
  Antonin Berthon (ENS & RIKEN AIP)+
  Bo Han (RIKEN AIP) Gang Niu (RIKEN AIP)
  Tongliang Liu (The University of Sydney)
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo)
• Learning Noise Transition Matrix from Only Noisy Labels via Total Variation Regularization
  Yivan Zhang (The University of Tokyo / RIKEN AIP)
  Gang Niu (RIKEN AIP)
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo)

Short talk (24 papers)

• Active Learning for Distributionally Robust Level-Set Estimation.
  Yu Inatsu (Nagoya Institute of Technology)
  Shogo Iwazaki (Nagoya Institute of Technology)
  Ichiro Takeuchi (Nagoya Institute of Technology / RIKEN AIP)
• Bias-Variance Reduced Local SGD for Less Heterogeneous Federated Learning
  Tomoya Murata (NTT DATA Mathematical Systems Inc. / The University of Tokyo)
  Taiji Suzuki (The University of Tokyo / RIKEN AIP)

• Binary Classification from Multiple Unlabeled Datasets via Surrogate Set Classification
  Shida Lei (The University of Tokyo)
  Nan Lu (The University of Tokyo/RIKEN AIP)
  Gang Niu (RIKEN AIP)
  Issei Sato (The university of Tokyo/RIKEN)
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo)
• CIFS: Improving Adversarial Robustness of CNNs via Channel-wise Importance-based Feature Selection Hanshu Yan (NUS)
  Jingfeng Zhang (RIKEN AIP)
  Gang Niu (RIKEN AIP)
  Jiashi Feng (NUS)
  Vincent Tan (NUS)
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo)
• Classification with Rejection Based on Cost-sensitive Classification
  Nontawat Charoenphakdee (The University of Tokyo / RIKEN AIP)
  Zhenghang Cui (The University of Tokyo / RIKEN AIP)
  Yivan Zhang (The University of Tokyo / RIKEN AIP)
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo)
• Class2Simi: A Noise Reduction Perspective on Learning with Noisy Labels
  Songhua Wu ( The University of Sydney )
  Xiaobo Xia ( The University of Sydney )
  Tongliang Liu ( The University of Sydney )
  Bo Han ( HKBU / RIKEN )
  Mingming Gong ( University of Melbourne )
  Nannan Wang ( Xidian University )
  Haifeng Liu ( Brain-Inspired Technology Co., Ltd. )
  Gang Niu ( RIKEN )
• Large-Margin Contrastive Learning with Distance Polarization Regularizer
  Shuo Chen (RIKEN AIP)
  Gang Niu (RIKEN AIP)
  Chen Gong (Nanjing University of Science and Technology )
  Jun Li (Nanjing University of Science and Technology )
  Jian Yang (Nanjing University of Science and Technology )
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo )
• Learning Diverse-Structured Networks for Adversarial Robustness
  Xuefeng Du (Unversity of Wisconsin, Madison)
  Jingfeng Zhang (RIKEN AIP)
  Bo Han (RIKEN AIP)
  Tongliang Liu (The University of Sydney)
  Yu Rong (Tencent AI Lab)
  Gang Niu (RIKEN AIP)
  Junzhou Huang (University of Texas at Arlington)
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo)
• Learning from Similarity-Confidence Data
  Yuzhou Cao (China Agricultural University)
  Lei Feng (Chongqing University)+
  Yitian Xu (China Agricultural University)
  Bo An (Nanyang Technological University)
  Gang Niu (RIKEN AIP)
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo)
• Lower-bounded proper losses for weakly supervised classification
  Shuhei Yoshida (NEC / RIKEN AIP)
  Takashi Takenouchi (Future University Hakodate/RIKEN AIP)
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo)
• Maximum Mean Discrepancy is Aware of Adversarial Attacks
  Ruize Gao (Hong Kong Baptist University )
  Feng Liu (University of Technology Sydney)+
  Jingfeng Zhang (RIKEN AIP)
  Bo Han (RIKEN AIP)
  Tongliang Liu ( The University of Sydney )
  Gang Niu (RIKEN AIP)
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo)
• Mediated uncoupled learning: Learning functions without direct input-output correspondences
  Ikko Yamane (Université Paris Dauphine – PSL/RIKEN AIP)
  Junya Honda (Kyoto University / RIKEN AIP)
  Florian Yger (Université Paris-Dauphine )
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo )
• More Powerful and General Selective Inference for Stepwise Feature Selection using Homotopy Method.
  Kazuya Sugiyama (Nagoya Institute of Technology)
  Vo Nguyen Le Duy (Nagoya Institute of Technology / RIKEN AIP)
  Ichiro Takeuchi (Nagoya Institute of Technology / RIKEN AIP)
• Non-exponentially weighted aggregation: regret bounds for unbounded loss functions
  Pierre Alquier (RIKEN AIP)
• On Learnability via Gradient Method for Two-Layer ReLU Neural Networks in Teacher-Student Setting
  Shunta Akiyama (The University of Tokyo)
  Taiji Suzuki (The University of Tokyo / RIKEN AIP)
• Optimal Transport Kernels for Sequential and Parallel Neural Architecture Search
  Vu Nguyen (Amazon)*
  Tam Le (RIKEN AIP)*
  Makoto Yamada (Kyoto University/RIKEN AIP)
  Michael A Osborne (The University of Oxford
  (*: equal contribution)
• Pointwise Binary Classification with Pairwise Confidence Comparisons
  Lei Feng (Chongqing University)+
  Senlin Shu (Southwest University)
  Nan Lu (The University of Tokyo/RIKEN)
  Bo Han (RIKEN AIP)
  Miao Xu (University of Queensland)
  Gang Niu (RIKEN AIP)
  Bo An (Nanyang Technological University)
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo)
• Positive-Negative Momentum: Manipulating Stochastic Gradient Noise to Improve Generalization
  Zeke Xie (The University of Tokyo/RIKEN AIP)
  Li Yuan (National Univerisity of Singapore)
  Zhanxing Zhu (Peking University)
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo)
• Post-selection inference with HSIC-Lasso
  Tobias Freidling (University of Cambridge)
  Benjamin Poignard (Osaka University/RIKEN AIP)
  Héctor Climente-González (RIKEN AIP)
  Makoto Yamada (Kyoto University/RIKEN AIP
• Provably End-to-end Label-noise Learning without Anchor Points
  Xuefeng Li (University of New South Wales)
  Tongliang Liu (The University of Sydney)
  Bo Han (RIKEN AIP)
  Gang Niu (RIKEN AIP)
  Masashi Sugiyama (RIKEN AIP/The University of Tokyo )
• Quantitative Understanding of VAE as a Non-linearly Scaled Isometric Embedding
  Akira Nakagawa (Fujitsu Laboratories Ltd.)
  Keizo Kato (Fujitsu Laboratories Ltd.)
  Taiji Suzuki (The University of Tokyo / RIKEN AIP)
• Scalable Marginal Likelihood Estimation for Model Selection in Deep Learning
  Alexander Immer (ETH and Max-Planck)+
  Matthias Bauer (Max-Planck and University of Cambridge)+
  Vincent Fortuin (ETH)
  Gunnar Rätsch (ETH and Max-Planck)
  Mohammad Emtiyaz Khan (RIKEN AIP)

• Supervised Tree-Wasserstein Distance
  Yuki Takezawa (Kyoto University/RIKEN AIP)
  Ryoma Sato (Kyoto University/RIKEN AIP)
  Makoto Yamada (Kyoto University/RIKEN AIP)

• Tractable structured natural gradient descent using local parameterizations
  Wu Lin (UBC)
  Frank Nielsen (Sony)
  Mohammad Emtiyaz Khan (RIKEN AIP)
  Mark Schmidt (UBC)

+Past interns of RIKEN AIP

関連研究室