研究者詳細

顔写真

スン　ル
孫　露
Lu Sun
所属
未踏スケールデータアナリティクスセンター　データアナリティクス研究部門
職名
准教授
学位

  • 博士（工学）(北海道大学)

  • 東北師範大学修士（理学）

e-Rad 研究者番号
11027195

経歴 5

  • 2026年5月 ～ 継続中
    特定国立研究開発法人理化学研究所 革新知能統合研究センター　客員研究員

  • 2026年2月 ～ 継続中
    東北大学　未踏スケールデータアナリティクスセンター　准教授

  • 2024年11月 ～ 2026年1月
    特定国立研究開発法人理化学研究所　革新知能統合研究センター　研究員

  • 2019年11月 ～ 2024年9月
    上海科技大学　情報科学技術研究科　助教授

  • 2017年10月 ～ 2019年10月
    京都大学　化学研究所　特定研究員

学歴 3

  • 北海道大学　大学院情報科学研究科

    2014年4月 ～ 2017年9月

  • 東北師範大学　大学院修士課程　回路とシステム

    2010年9月 ～ 2013年6月

  • 武漢大学　情報理工学部

    2005年9月 ～ 2009年6月

研究キーワード 1

  • 解釈可能機械学習, 特徴選択, モデル圧縮

研究分野 1

  • 情報通信 / 知能情報学 / 人工知能, 機械学習, 深層学習

論文 36

  1. Learning Local Feature Masks with Variational Information Bottleneck 査読有り

    Lu Sun, Jun Sakuma

    Proceedings of the 35th International Joint Conference on Artificial Intelligence (IJCAI-ECAI 2026)　2026年8月

  2. Learning Semi-Structured Sparsity for LLMs via Shared and Context-Aware Hypernetwork 査読有り

    Lu Sun, Jun Sakuma

    Proceedings of The 14th International Conference on Learning Representations (ICLR 2026)　2026年4月

  3. Multi-level sparse network lasso: Locally sparse learning with flexible sample clusters 査読有り

    Luhuan Fei, Xinyi Wang, Jiankun Wang, Lu Sun, Yuyao Zhang

    Neurocomputing　635　129898-129898　2025年6月

    出版者・発行元： Elsevier BV

    DOI： 10.1016/j.neucom.2025.129898 　

    ISSN：0925-2312

  4. Multi-view multi-label personalized classification via generalized exclusive sparse tensor factorization 査読有り

    Luhuan Fei, Weijia Lin, Jiankun Wang, Lu Sun, Mineichi Kudo, Keigo Kimura

    Knowledge and Information Systems　67　(9)　8023-8057　2025年5月29日

    出版者・発行元： Springer Science and Business Media LLC

    DOI： 10.1007/s10115-025-02449-8 　

    ISSN：0219-1377

    eISSN：0219-3116

  5. Multi-level network Lasso for multi-task personalized learning 査読有り

    Jiankun Wang, Luhuan Fei, Lu Sun

    Pattern Recognition　161　111213-111213　2025年5月

    出版者・発行元： Elsevier BV

    DOI： 10.1016/j.patcog.2024.111213 　

    ISSN：0031-3203

  6. Robust multi-view subspace clustering with missing data by aligning nonlinear manifolds 査読有り

    Zhan-Wang Mao, Lu Sun, Youlong Wu

    Pattern Recognition　2025年5月

    DOI： 10.1016/j.patcog.2024.111280 　

  7. Learning Compact Neural Networks via Generalized Structured Sparsity 査読有り

    Ke Bian, Lu Sun, Dengji Zhao

    Frontiers in Artificial Intelligence and Applications　2024年10月16日

    出版者・発行元： IOS Press

    DOI： 10.3233/faia240791 　

    ISSN：1879-8314

  8. Learning Low-Rank Tensor Cores with Probabilistic ℓ0-Regularized Rank Selection for Model Compression 査読有り

    Sylvain Bouveret, Hugo Gilbert, Jérôme Lang, Guillaume Méroué

    Proceedings of the Thirty-ThirdInternational Joint Conference on Artificial Intelligence　3762-3770　2024年8月

    出版者・発行元： International Joint Conferences on Artificial Intelligence Organization

    DOI： 10.24963/ijcai.2024/418 　

    詳細を見る 詳細を閉じる

    When allocating indivisible items to agents, it is known that the only strategyproof mechanisms that satisfy a set of rather mild conditions are constrained serial dictatorships: given a fixed order over agents, at each step the designated agent chooses a given number of items (depending on her position in the sequence). Agents who come earlier in the sequence have a larger choice of items; however, this advantage can be compensated by a higher number of items received by those who come later. How to balance priority in the sequence and number of items received is a nontrivial question. We use a previous model, parameterized by a mapping from ranks to scores, a social welfare functional, and a distribution over preference profiles. For several meaningful choices of parameters, we show that the optimal sequence can be computed exactly in polynomial time or approximated using sampling. Our results hold for several probabilistic models on preference profiles, with an emphasis on the Plackett-Luce model. We conclude with experimental results showing how the optimal sequence is impacted by various parameters.

  9. MoME: Mixture-of-Masked-Experts for Efficient Multi-Task Recommendation 査読有り

    Jiahui Xu, Lu Sun, Dengji Zhao

    Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval　2527-2531　2024年7月10日

    出版者・発行元： ACM

    DOI： 10.1145/3626772.3657922 　

  10. Redirected transfer learning for robust multi-layer subspace learning 査読有り

    Jiaqi Bao, Mineichi Kudo, Keigo Kimura, Lu Sun

    Pattern Analysis and Applications　2024年3月

    DOI： 10.1007/s10044-024-01233-8 　

  11. Robust embedding regression for semi-supervised learning 査読有り

    Bao, J., Kudo, M., Kimura, K., Sun, L.

    Pattern Recognition　145　2024年

    DOI： 10.1016/j.patcog.2023.109894 　

    ISSN：0031-3203

  12. Incomplete Multi-view Weak-Label Learning with Noisy Features and Imbalanced Labels 査読有り

    Zhiwei Li, Zijian Yang, Lu Sun, Mineichi Kudo, Keigo Kimura

    PRICAI 2023: Trends in Artificial Intelligence　2024年

    DOI： 10.1007/978-981-99-7022-3_12 　

    ISSN：0302-9743 1611-3349

  13. Partial Multi-label Learning with a Few Accurately Labeled Data 査読有り

    Haruhi Mizuguchi, Keigo Kimura, Mineichi Kudo, Lu Sun

    PRICAI 2023: Trends in Artificial Intelligence　2024年

    DOI： 10.1007/978-981-99-7022-3_7 　

    ISSN：0302-9743 1611-3349

  14. Multi-Label Personalized Classification via Exclusive Sparse Tensor Factorization 査読有り

    Weijia Lin, Jiankun Wang, Lu Sun, Mineichi Kudo, Keigo Kimura

    2023 IEEE International Conference on Data Mining (ICDM)　398-407　2023年12月1日

    出版者・発行元： IEEE

    DOI： 10.1109/icdm58522.2023.00049 　

  15. Generalized Discriminative Deep Non-Negative Matrix Factorization Based on Latent Feature and Basis Learning 査読有り

    Zijian Yang, Zhiwei Li, Lu Sun

    Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence　4486-4494　2023年8月

    出版者・発行元： International Joint Conferences on Artificial Intelligence Organization

    DOI： 10.24963/ijcai.2023/499 　

    詳細を見る 詳細を閉じる

    As a powerful tool for data representation, deep NMF has attracted much attention in recent years. Current deep NMF builds the multi-layer structure by decomposing either basis matrix or feature matrix into multiple factors, and probably complicates the learning process when data is insufficient or exhibits simple structure. To overcome the limitations, a novel method called Generalized Deep Non-negative Matrix Factorization (GDNMF) is proposed, which generalizes several NMF and deep NMF methods in a unified framework. GDNMF simultaneously performs decomposition on both features and bases, which learns a hierarchical data representation based on multi-level basis. To further improve the latent representation and enhance its flexibility, GDNMF mutually reinforces shallow linear model and deep non-linear model. Moreover, semi-supervised GDNMF is proposed by treating partial label information as soft constraints in the multi-layer structure. An efficient two-phase optimization algorithm is developed, and experiments on five real-world datesets verify its superior performance compared with state-of-the-art methods.

  16. Fed-SC: One-Shot Federated Subspace Clustering over High-Dimensional Data 査読有り

    Songjie Xie, Youlong Wu, Kewen Liao, Lu Chen, Chengfei Liu, Haifeng Shen, MingJian Tang, Lu Sun

    2023 IEEE 39th International Conference on Data Engineering (ICDE)　2905-2918　2023年4月

    出版者・発行元： IEEE

    DOI： 10.1109/icde55515.2023.00222 　

  17. Structured Sparse Multi-Task Learning with Generalized Group Lasso 査読有り

    Fei, L., Sun, L., Kudo, M., Kimura, K.

    Frontiers in Artificial Intelligence and Applications　372　2023年

    出版者・発行元： Frontiers in Artificial Intelligence and Applications

    DOI： 10.3233/FAIA230333 　

    ISSN：0922-6389

  18. Grouped Multi-Task Learning with Hidden Tasks Enhancement 査読有り

    Jin, J., Wang, J., Sun, L., Zheng, J., Kudo, M.

    Frontiers in Artificial Intelligence and Applications　372　2023年

    出版者・発行元： Frontiers in Artificial Intelligence and Applications

    DOI： 10.3233/FAIA230392 　

    ISSN：0922-6389

  19. Multiplicative Sparse Tensor Factorization for Multi-View Multi-Task Learning 査読有り

    Wang, X., Sun, L., Nguyen, C.H., Mamitsuka, H.

    Frontiers in Artificial Intelligence and Applications　372　2023年

    出版者・発行元： Frontiers in Artificial Intelligence and Applications

    DOI： 10.3233/FAIA230561 　

    ISSN：0922-6389

  20. Multi-Task Personalized Learning with Sparse Network Lasso 査読有り

    Jiankun Wang, Lu Sun

    Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence　3516-3522　2022年7月

    出版者・発行元： International Joint Conferences on Artificial Intelligence Organization

    DOI： 10.24963/ijcai.2022/488 　

    詳細を見る 詳細を閉じる

    Multi-task learning learns multiple related tasks together, in order to improve the generalization performance. Existing methods typically build a global model shared by all the samples, which saves the homogeneity but ignores the individuality (heterogeneity) of samples. Personalized learning is recently proposed to learn sample-specific local models by utilizing sample heterogeneity, however, directly applying it in the multi-task learning setting poses three key challenges: 1) model sample homogeneity, 2) prevent from over-parameterization and 3) capture task correlations. In this paper, we propose a novel multi-task personalized learning method to handle these challenges. For 1), each model is decomposed into a sum of global and local components, that saves sample homogeneity and sample heterogeneity, respectively. For 2), regularized by sparse network Lasso, the joint models are embedded into a low-dimensional subspace and exhibit sparse group structures, leading to a significantly reduced number of effective parameters. For 3), the subspace is further separated into two parts, so as to save both commonality and specificity of tasks. We develop an alternating algorithm to solve the proposed optimization problem, and extensive experiments on various synthetic and real-world datasets demonstrate its robustness and effectiveness.

  21. Efficient Leave-One-Out Evaluation of Kernelized Implicit Mappings 査読有り

    Kudo, M., Kimura, K., Morishita, S., Sun, L.

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)　13813 LNCS　2022年

    出版者・発行元： Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

    DOI： 10.1007/978-3-031-23028-8_23 　

    ISSN：1611-3349 0302-9743

  22. Realization of Autoencoders by Kernel Methods 査読有り

    Morishita, S., Kudo, M., Kimura, K., Sun, L.

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)　13813 LNCS　2022年

    出版者・発行元： Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

    DOI： 10.1007/978-3-031-23028-8_1 　

    ISSN：1611-3349 0302-9743

  23. Retargeted Regression Methods for Multi-label Learning 査読有り

    Kimura, K., Bao, J., Kudo, M., Sun, L.

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)　13813 LNCS　2022年

    出版者・発行元： Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

    DOI： 10.1007/978-3-031-23028-8_21 　

    ISSN：1611-3349 0302-9743

  24. Multiplicative Sparse Feature Decomposition for Efficient Multi-View Multi-Task Learning 査読有り

    Lu Sun, Canh Hao Nguyen, Hiroshi Mamitsuka

    Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence　3506-3512　2019年8月

    出版者・発行元： International Joint Conferences on Artificial Intelligence Organization

    DOI： 10.24963/ijcai.2019/486 　

    詳細を見る 詳細を閉じる

    Multi-view multi-task learning refers to dealing with dual-heterogeneous data,where each sample has multi-view features,and multiple tasks are correlated via common views.Existing methods do not sufficiently address three key challenges:(a) saving task correlation efficiently, (b) building a sparse model and (c) learning view-wise weights.In this paper, we propose a new method to directly handle these challenges based on multiplicative sparse feature decomposition.For (a), the weight matrix is decomposed into two components via low-rank constraint matrix factorization, which saves task correlation by learning a reduced number of model parameters.For (b) and (c), the first component is further decomposed into two sub-components,to select topic-specific features and learn view-wise importance, respectively. Theoretical analysis reveals its equivalence with a general form of joint regularization,and motivates us to develop a fast optimization algorithm in a linear complexity w.r.t. the data size.Extensive experiments on both simulated and real-world datasets validate its efficiency.

  25. Fast and Robust Multi-View Multi-Task Learning via Group Sparsity 査読有り

    Lu Sun, Canh Hao Nguyen, Hiroshi Mamitsuka

    Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence　3499-3505　2019年8月

    出版者・発行元： International Joint Conferences on Artificial Intelligence Organization

    DOI： 10.24963/ijcai.2019/485 　

    詳細を見る 詳細を閉じる

    Multi-view multi-task learning has recently attracted more and more attention due to its dual-heterogeneity, i.e.,each task has heterogeneous features from multiple views, and probably correlates with other tasks via common views.Existing methods usually suffer from three problems: 1) lack the ability to eliminate noisy features, 2) hold a strict assumption on view consistency and 3) ignore the possible existence of task-view outliers.To overcome these limitations, we propose a robust method with joint group-sparsity by decomposing feature parameters into a sum of two components,in which one saves relevant features (for Problem 1) and flexible view consistency (for Problem 2),while the other detects task-view outliers (for Problem 3).With a global convergence property, we develop a fast algorithm to solve the optimization problem in a linear time complexity w.r.t. the number of features and labeled samples.Extensive experiments on various synthetic and real-world datasets demonstrate its effectiveness.

  26. Multi-label classification by polytree-augmented classifier chains with label-dependent features 査読有り

    Sun, L., Kudo, M.

    Pattern Analysis and Applications　22　(3)　2019年

    DOI： 10.1007/s10044-018-0711-6 　

  27. Optimization of classifier chains via conditional likelihood maximization 査読有り

    Sun, L., Kudo, M.

    Pattern Recognition　74　2018年

    DOI： 10.1016/j.patcog.2017.09.034 　

  28. READER: Robust semi-supervised multi-label dimension reduction 査読有り

    Sun, L., Kudo, M., Kimura, K.

    IEICE Transactions on Information and Systems　E100D　(10)　2017年

    DOI： 10.1587/transinf.2017EDP7184 　

  29. Multi-label classification with meta-label-specific features 査読有り

    Lu Sun, Mineichi Kudo, Keigo Kimura

    2016 23rd International Conference on Pattern Recognition (ICPR)　1612-1617　2016年12月

    出版者・発行元： IEEE

    DOI： 10.1109/icpr.2016.7899867 　

  30. Locality in multi-label classification problems 査読有り

    Batzaya Norov-Erdene, Mineichi Kudo, Lu Sun, Keigo Kimura

    2016 23rd International Conference on Pattern Recognition (ICPR)　2319-2324　2016年12月

    出版者・発行元： IEEE

    DOI： 10.1109/icpr.2016.7899982 　

  31. Fast random k-labELsets for large-scale multi-label classification 査読有り

    Keigo Kimura, Mineichi Kudo, Lu Sun, Sadamori Koujaku

    2016 23rd International Conference on Pattern Recognition (ICPR)　438-443　2016年12月

    出版者・発行元： IEEE

    DOI： 10.1109/icpr.2016.7899673 　

  32. A scalable clustering-based local multi-label classification method 査読有り

    Sun, L., Kudo, M., Kimura, K.

    Frontiers in Artificial Intelligence and Applications　285　2016年

    出版者・発行元： Frontiers in Artificial Intelligence and Applications

    DOI： 10.3233/978-1-61499-672-9-261 　

  33. Simultaneous nonlinear label-instance embedding for multi-label classification 査読有り

    Kimura, K., Kudo, M., Sun, L.

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)　10029 LNCS　2016年

    出版者・発行元： Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

    DOI： 10.1007/978-3-319-49055-7_2 　

  34. Dimension Reduction Using Nonnegative Matrix Tri- Factorization in Multi-label Classification 査読有り

    Keigo Kimura, Mineichi Kudo, Lu Sun

    Proceedings of the 21st International Conference on Parallel & Distributed Processing Techniques & Applications: Workshop on Mathematical Modeling and Problem Solving (PDPTA 2015)　250-255　2015年8月

  35. Polytree-Augmented Classifier Chains for Multi-Label Classification 査読有り

    Lu Sun, Mineichi Kudo

    Proceedings of the 24th International Joint Conference of Artificial Intelligence (IJCAI 2015)　3834-3840　2015年8月

  36. Novel fast license plate location method based on hierarchical strategy

    Huang, J.-P., Sun, L., Qiao, S., Sun, J.-N.

    Journal of Jilin University (Engineering and Technology Edition)　45　(2)　2015年

    DOI： 10.13229/j.cnki.jdxbgxb201502045 　

︎全件表示 ︎最初の5件までを表示

共同研究・競争的資金等の研究課題 2

  1. 効率的・信頼可能・転移可能な次世代AIのための統一的モデル圧縮

    提供機関：Japan Society for the Promotion of Science

    研究種目：FY2026 Early-Career Research

    2026年4月 ～ 2028年3月

  2. ゲノム・化学相互作用の解読に向けた解釈可能なグラフ機械学習

    提供機関：Institute for Chemical Research, Kyoto University

    制度名：FY2026 International Joint Usage / Joint Research

    2026年4月 ～ 2027年3月