Kohei Hatano | Last modified date：2022.12.08 |

Graduate School

Undergraduate School

E-Mail *Since the e-mail address is not displayed in Internet Explorer, please use another web browser:Google Chrome, safari.

Homepage

##### https://kyushu-u.pure.elsevier.com/en/persons/kohei-hatano

Reseacher Profiling Tool Kyushu University Pure

##### https://sites.google.com/view/koheihatano/home

Phone

092-802-6031

Academic Degree

Ph.D

Country of degree conferring institution (Overseas)

No

Field of Specialization

machine learning

ORCID(Open Researcher and Contributor ID)

0000-0002-1536-1269

Total Priod of education and research career in the foreign country

00years03months

Outline Activities

My main research interest includes Machine Learning, in particular, design and analyses of robust and efficient machine learning algorithms from a theoretical perspective. I am also affiliated with Department of Informatics.

Research

**Research Interests**

- Theory of machine learning and development of algorithms

keyword : machine learning, computational learning theory, boosting, optimization

2000.04[Theoretical analyses of boosting methods] Boosting is a technique for constructing a highly accurate classifier by combining many ``weak'' classifiers. Although boosting has become a fundamental tool in machine learning and data mining these days, its theoretical propeties are yet to be understood. We have analysed theoretical properties of boosting in order to design more efficient boosting methods. As a result, we clarified a property which might be a key for further improvement. So far, we are developping a new boosting algorithm based on our analysis.. - Design and analysis of online prediction algorithms

keyword : online learning, computational learning theory, machine learning, optimization, ranking

2007.05. - learning analytics

keyword : educational data, learning analytics

2017.11. - Open Science

keyword : open access, open data, open science

2015.08～2017.10.

**Academic Activities**

**Reports**

1. | , [URL]. |

2. | Report on The 21st Annual Conference on Learning Theory (COLT 2008) . |

3. | Frontier of Theoretical Computer Science. |

**Papers**

1. | Sherief Hashima, Mostafa M. Fouda, Zubair Md Fadlullah, Ehab Mahmoud Mohamed, Kohei Hatano, Improved UCB-based Energy-Efficient Channel Selection in Hybrid-Band Wireless Communication, Proceedings of GLOBECOM '21, 10.1109/GLOBECOM46510.2021.9685996, 2021.12. |

2. | Yaxiong Liu, Ken-ichiro Moridomi, Kohei Hatano, Eiji Takimoto, An online semi-definite programming with a generalised log-determinant regularizer and its applications, Proceedings of The 13th Asian Conference on Machine Learning (ACML), PMLR 157, 1113-1128, 2021.11. |

3. | Yaxiong Liu, Xuanke Jiang, Kohei Hatano, Eiji Takimoto, Expert advice problem with noisy low rank loss, Proceedings of The 13th Asian Conference on Machine Learning (ACML), PMLR 157, 1097-1112, 2021.11. |

4. | Ehab Mahmoud Mohamed, Sherief Hashima, Kohei Hatano, Mostafa M. Fouda, Zubair Md. Fadlullah, Sleeping Contextual/Non-Contextual Thompson Sampling MAB for mmWave D2D Two-Hop Relay Probing., IEEE Transactions on Vehicular Technology, 10.1109/TVT.2021.3116223, 70, 11, 12101-12112, 2021.11. |

5. | Fumito Miyake, Eiji Takimoto, kohei hatano, Succinct representation of linear extensions via MDDs and its application to scheduling under precedence constraints, 30th International Workshop on Combinatorial Algorithms, IWOCA 2019
Combinatorial Algorithms - 30th International Workshop, IWOCA 2019, Proceedings, 10.1007/978-3-030-25005-8_30, 365-377, 2019.07, We consider a single machine scheduling problem to minimize total flow time under precedence constraints, which is NP-hard. Matsumoto et al. proposed an exact algorithm that consists of two phases: first construct a Multi-valued Decision Diagram (MDD) to represent feasible permutations of jobs, and then find the shortest path in the MDD which corresponds to the optimal solution. Although their algorithm performs significantly better than standard IP solvers for problems with dense constraints, the performance rapidly diminishes when the number of constraints decreases, which is due to the exponential growth of MDDs. In this paper, we introduce an equivalence relation among feasible permutations and show that it suffices to construct an MDD that maintains only one representative for each equivalence class. Experimental results show that our algorithm outperforms Matsumoto et al.’s algorithm for problems with sparse constraints, while keeping good performance for dense constraints. Moreover, we show that Matsumoto et al.’s algorithm can be extended for solving a more general problem of minimizing weighted total flow time.. |

6. | Kohei Hatano, Combinatorial Online Prediction, 15th International Symposium on Information Theory and Its Applications, ISITA 2018
Proceedings of 2018 International Symposium on Information Theory and Its Applications, ISITA 2018, 10.23919/ISITA.2018.8664224, 40-44, 2019.03, We present a short survey on recent results on combinatorial online prediction in the adversarial setting.. |

7. | Takahiro Fujita, Kohei Hatano, Eiji Takimoto, Boosting over non-deterministic ZDDs, Theoretical Computer Science, https://doi.org/10.1016/j.tcs.2018.11.027, 2018.12, [URL], 本論文では，ZDDと呼ばれる圧縮データ構造を用いてデータを圧縮した上で，圧縮したデータ上でブースティングという機械学習手法が効率よく計算できる事を明らかにした．ビッグデータの解析において，省スペースで計算が出来ることは大きな利点である．本研究の興味深い点は，組合せ論的オンライン予測という（一見）全く異なる機械学習分野の知見が圧縮情報処理に活かされたことである．. |

8. | Takahiro Fujita, Kohei Hatano, and Eiji Takimoto, “, Online Combinatorial Optimization with Multiple Projections and Its Application to Scheduling Problem Volume and Number: Vol.,pp.-,Sep. 2018., IEICE Transactions on Information and Systems, E101-A, 9, 2018.09. |

9. | Takahiro Fujita, Kohei Hatano, Shuji Kijima, Eiji Takimoto, Online combinatorial optimization with multiple projections and its application to scheduling problem, IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences, 10.1587/transfun.E101.A.1334, E101A, 9, 1334-1343, 2018.09, We consider combinatorial online prediction problems and propose a new construction method of efficient algorithms for the problems. One of the previous approaches to the problem is to apply online prediction method, in which two external procedures the projection and the metarounding are assumed to be implemented. In this work, we generalize the projection to multiple projections. As an application of our framework, we show an algorithm for an online job scheduling problem with a single machine with precedence constraints.. |

10. | Ken-ichiro Moridomi, Kohei Hatano, Eiji Takimoto, Tighter generalization bounds for matrix completion via factorization into constrained matrices, IEICE Transactions on Information and Systems, 10.1587/transinf.2017EDP7339, E101D, 8, 1997-2004, 2018.08, We prove generalization error bounds of classes of low-rank matrices with some norm constraints for collaborative filtering tasks. Our bounds are tighter, compared to known bounds using rank or the related quantity only, by taking the additional L_{1} and L_{∞} constraints into account. Also, we show that our bounds on the Rademacher complexity of the classes are optimal.. |

11. | Ken-ichiro Moridomi, Kohei Hatano, and Eiji Takimoto, “, Online linear optimization with the log-determinant regularizer, IEICE Transactions on Information and Systems, E101-D, 6, 2018.06. |

12. | Daiki Suehiro, Kohei hatano, Eiji Takimoto, Efficient reformulation of 1-norm ranking SVM, IEICE Transactions on Information and Systems, 10.1587/transinf.2017EDP7233, E101D, 3, 719-729, 2018.03. |

13. | Takahiro Fujita, Kohei Hatano, Eiji Takimoto, Boosting over non-deterministic ZDDs, Proceedings of the 12th International Conference and Workshop on Algorithms and Computation( WALCOM 2018), 10.1007/978-3-319-75172-6_17, 195-206, 2018.01, 本論文では，ZDDと呼ばれる圧縮データ構造を用いてデータを圧縮した上で，圧縮したデータ上でブースティングという機械学習手法が効率よく計算できる事を明らかにした．ビッグデータの解析において，省スペースで計算が出来ることは大きな利点である．本研究の興味深い点は，組合せ論的オンライン予測という（一見）全く異なる機械学習分野の知見が圧縮情報処理に活かされたことである．. |

14. | Kohei Hatano, Can machine learning techniques provide better learning support for elderly people?, 6th International Conference on Distributed, Ambient and Pervasive Interactions, DAPI 2018 Held as Part of HCI International 2018
Distributed, Ambient and Pervasive Interactions
Technologies and Contexts - 6th International Conference, DAPI 2018, Held as Part of HCI International 2018, Proceedings, 10.1007/978-3-319-91131-1_14, 178-187, 2018.01, Computer-based support for learning of elderly people is now considered as an important issue in the super-aged society. Extra cares are needed for elderly people’s learning compared to younger people, since they might have difficulty in using computers, reduced cognitive ability and other physical problems which make them less motivated. Key components of a better learning support system are sensing the contexts surrounding elderly people and providing appropriate feedbacks to them. In this paper, we review some existing techniques of the contextual bandit framework in the machine learning literature, which could be potentially useful for online decision making scenarios given contexts. We also discuss issues and challenges to apply the framework.. |

15. | Nir Ailon, Kohei Hatano, Eiji Takimoto, Bandit Online Optimization Over the Permutahedron, Theoretical Computer Science, 10.1016/j.tcs.2016.07.033, 650, 18, 92-108, 2016.10. |

16. | Takumi Nakazono, Ken-ichiro Moridomi, Kohei Hatano, Eiji Takimoto, A Combinatorial Metrical Task System Problem under the Uniform Metric, Proceedings of 27th International Conference on Algorithmic Learning Theory(ALT 2016), 10.1007/978-3-319-46379-7_19, LNCS 9926, 276-287, 2016.10. |

17. | Takahiro Fujita, Kohei Hatano, Shuji Kijima, Eiji Takimoto, Online Linear Optimization for Job Scheduling under Precedence Concstraints, Proceedings of 26th International Conference on Algorithmic Learning Theory(ALT 2015), 10.1007/978-3-319-24486-0_22, LNCS 6331, 345-359, 2015.10. |

18. | Issei Matsumoto, Kohei Hatano, Eiji Takimoto, Online Density Estimation of Bradley-Terry Models, Proceedings of the 28th Conference on Learning Theory (COLT 2015), JMLR W&CP 40, 1343-1359, 2015.06. |

19. | Nir Ailon, Kohei Hatano, Eiji Takimoto, Bandit Online Optimization Over the Permutahedron, Proceedings of the 25th International Conference on Algorithmic Learning Theory (ALT 2014),, LNCS 8776, 215–229, 2014.10. |

20. | Kazuki Teraoka, Kohei Hatano, Eiji Takimoto, Efficient Sampling Method for Monte Carlo Tree Search, IEICE TRANSACTIONS on Information and System, E97-D, 3, 392-298, 2014.03. |

21. | Takahiro Fujita, Kohei Hatano, Eiji Takimoto, Combinatorial Online Prediction via Metarounding, Proceedings of the 24th International Conference on Algorithmic Learning Theory (ALT 2013), 68-82, 2013.10. |

22. | Shota Yasutake, Kohei Hatano, Eiji Takimoto, Masayuki Takeda, Online Rank Aggregation , Proceedings of the 4th Asian Conference on Machine Learning(ACML 2012)
, 539-553, 2012.11. |

23. | Yoko Anan, Kohei Hatano, Hideo Bannai, Masayuki Takeda, Polyphonic Music Classification on Symbolic Data Using Dissimilarity Functions, Proceedings of the 13th International Society for Music Information Retrieval Conference (ISMIR 2012), 229-234, 2012.10. |

24. | Daiki Suehiro, Kohei Hatano, Shuji Kijima, Eiji Takimoto, Kiyohito Nagano, Online Prediction under Submodular Constraints, Proceedings of the 23rd International Conference on Algorithmic Learning Theory (ALT 2012)
, 260-274, 2012.10. |

25. | Shota Yasutake, Kohei Hatano, Shuji Kijima, Eiji Takimoto, Masayuki Takeda, , Online Linear Optimization over Permutations , Proceedings of the 22nd International Symposium on Algorithms and Computation (ISAAC 2011)
, 534-543, 2011.11. |

26. | Daiki Suehiro, Kohei Hatano, Eiji Takimoto, Approximate Reduction from AUC Maximization to 1-norm Soft Margin Optimization, Proceedings of the 22nd International Conference on Algorithmic Learning Theory (ALT 2011)
, 324-337, 2011.10. |

27. | Yoko Anan, Kohei Hatano, Hideo Bannai, Masayuki Takeda, Music Genre Classification using Similarity Functions, Proceedings of the 12th International Society for Music Information Retrieval Conference (ISMIR 2011), 693-698, 2011.10. |

28. | Shin-ichi Yoshida, Kohei Hatano, Eiji Takimoto, Masayuki Takeda, Adaptive Online Prediction Using Weighted Windows, IEICE Transactions on Information and Systems , E94-D, 10, 1917-1923, 2011.10. |

29. | Michinari Momma, Kohei Hatano, and Hiroki Nakayama, Ellipsoidal Support Vector Machines , Proceedings of the 2nd Asian Conference on Machine Learning (ACML 2010), 31-46, 2010.11. |

30. | Kazuaki Kashihara, Kohei Hatano, Hideo Bannnai, and Masayuki Takeda, Sparse Substring Pattern Set Discovery using Linear Programming Boosting, Proceedings of the 13th International Conference on Discovery Science (DS 2010), 132-143, 2010.10. |

31. | Kohei Hatano and Eiji Takimoto, Linear Programming Boosting by Column and Row Generation, Proceedings of the Twelfth International Conference on Discovery Science (DS'09) , 2009.10. |

32. | Jun-ichi Moribe, Kohei Hatano, Eiji Takimoto, and Masayuki Takeda, Smooth Boosting for Margin-Based Ranking, Proceedings of 19th International Conference on Algorithmic Learning Theory, 227-239, 2008.10. |

33. | Kosuke Ishibashi, Kohei Hatano, and Masayuki Takeda, Online Learning of Approximate Maximum p-Norm Margin Classifiers with Biases, Proceedings of the 21st Annual Conference on Learning Theory, 69—80, 2008.07. |

34. | Kosuke Ishibashi, Kohei Hatano, and Masayuki Takeda, Online Learning of Approximate Maximum Margin Classifiers with Biases, Proceedings of the 2nd International Workshop on Data Mining and Statistical Science, 2007.10. |

35. | Kohei Hatano, Smooth Boosting Using an Information-based Criterion, The 17 th international conference on algorithmic learning theory, 10.1007/11894841_25, 304-318, LNAI 4264., 2006.10. |

**Presentations**

1. | Kohei Hatano, Blackwell game and its applications in online prediction tasks, Forum "Math-for-Industry" 2021, 2021.12. |

2. | Yaxiong Liu, Kohei Hatano, and Eiji Takimoto, Improved Algorithms for Online Load Balancing, 情報処理学会 第174回アルゴリズム研究会, 2019.09. |

3. | , [URL]. |

4. | Takahiro Fujita, Kohei Hatano, Shuji Kijima, Eiji Takimoto, Online Linear Optimization over Permutations with Precedence Constraints , NIPS 2014 Workshop on Discrete Optimization in Machine Learning(DISCML), 2014.12. |

5. | Issei Matsumoto, Kohei Hatano, Eiji Takimoto, Online Prediction with Bradley-Terry Models, NIPS 2014 Workshop on Analysis of Rank Data: Confluence of Social Choice, Operations Research, and Machine Learning, 2014.12. |

6. | 畑埜 晃平, Combinatorial Online Prediction via Metarounding, TCE Guest Lecture, 2013.12. |

7. | Takahiro Fujita, Kohei Hatano, Eiji Takimoto, Combinatorial Online Prediction Using Offline Approximation Algorithms, The sixth Annual Meeting of Asian Association for Algorithms and Computation (AAAC2013), 2013.04. |

8. | Daiki Suehiro, Kohei Hatano, Shuji Kijima, Eiji Takimoto, Kiyohito Nagano, Online Prediction over Base Polyhedra , NIPS 2012 Workshop on Discrete Optimization in Machine Learning(DISCML), 2012.12. |

9. | Online Rank Aggregation. |

10. | Online Rank Aggregation. |

11. | Learning evaluation functions for Shogi via bipertite ranking learning with SVMs . |

12. | Learning evaluation functions for Shogi via bipertite ranking learning with SVMs . |

13. | Online Rank Aggregation. |

14. | Online Rank Aggregation. |

15. | Online Learning Based On Maximum Entropy Principle. |

16. | Online Learning of Maximum p-Norm Margin Classifiers with Bias. |

17. | Efficient online learning of linear threshold functions with large biases.. |

18. | Boosting using classifiers with nearly one-sided error. |

**Membership in Academic Society**

- EATCS
- IEICE
- LA

**Awards**

- We show that one of machine learning techniques called “boosting” can be efficiently computed over compressed data using non-determinist Zero-Suppressed Binary Decision Diagrams (ZDDs). A big advantage of this scheme is the ability to perform analysis using big data with minimal space. An interesting point in this research is that the combinatorial online prediction, a seemingly unrelated machine learning concept is successfully adopted for utilizing compressed data.
- JSAI Incentive Award
- JSAI Incentive Award

Educational

Unauthorized reprint of the contents of this database is prohibited.