PUBLICATIONS

List of all publication is here.

2023

Conferences

  1. Naoki Hayashi, Yoshihide Sawada. "Bayesian Generalization Error in Three-layered Linear Neural Network with Concept Bottleneck Structure and Comparison with Multi-task Formulation", JSAI2023, Kumamoto. (2023/6/8). in Japanese.
  2. Daigo Miyoshi, Sho Toyooka, Naoki Hayashi, Takumi Ichikawa, Shotaro Matsugi, Kotaro Ito, Tomoki Sekiguchi, Nobuyuki Ishikawa. "The development of delivery optimization algorithms for real-business applications of immediate and planned delivery", JSAI2023, Kumamoto. (2023/6/7). in Japanese.
  3. Naoki Hayashi, Yoshihide Sawada. "Real Log Canonical Threshold of Partial Concept Bottleneck Model and its Application to Bayesian Learning", IBIS2023, Kita-Kyushu. (2023/10/30). in Japanese.

Invited Talks

  1. Naoki Hayashi. "Singular Learning Theory of Interpretable Parameter Restriction", Developmental Interpretability Conference 2023 Nov., Online. (2023/11/11). Slides

2022

Conferences

  1. Naoki Hayashi, Yoshihide Sawada. "Effect of Concept Bottleneck Structure to Bayesian Generalization Error in Linear Neural Networks", IBIS2022, Tsukuba. (2022/11/21). in Japanese.

2021

Papers

  1. Naoki Hayashi. "The Exact Asymptotic Form of Bayesian Generalization Error in Latent Dirichlet Allocation", Neural Networks, Volume 137, March 2021, pp.127-137. doi: 10.1016/j.neunet.2021.01.024.
  2. Akira Endo, Mitsuo Uchida, Naoki Hayashi, Yang Liu, Katherine E. Atkins, Adam J. Kucharski, Sebastian Funk. "Within and between classroom transmission patterns of seasonal influenza among primary school students in Matsumoto city, Japan". Proceedings of the National Academy of Sciences (PNAS), Nov 2021, 118 (46) e2112605118. doi: 10.1073/pnas.2112605118

2020

Papers

  1. Naoki Hayashi, Sumio Watanabe. "Asymptotic Bayesian Generalization Error in Latent Dirichlet Allocation and Stochastic Matrix Factorization", SN Computer Science, Volume 1, 69 (2020), pp.1-22. doi: 10.1007/s42979-020-0071-3. (2019/8/24 submitted. 2020/1/30 accepted. 2020/2/20 published on web).
  2. Naoki Hayashi. "Variational Approximation Error in Non-negative Matrix Factorization", Neural Networks, Volume 126, June 2020, pp.65-75. doi: 10.1016/j.neunet.2020.03.009. (2019/6/18 submitted. 2020/3/9 accepted. 2020/3/20 published on web).
  3. Keita Harada, Naoki Hayashi, Katsushi Kagaya. "Individual behavioral type captured by a Bayesian model comparison of cap making by sponge crabs", PeerJ 8:e9036, pp.1-26.
    doi: 10.7717/peerj.9036. (Hayashi is added as an author in 2020/2/29, 2020/4/1 accepted, 2020/5/14 published).

International Conferences (reviewed)

  1. Akira Ito, Masaru Okutsu, Masatoshi Yukishima, Ryosuke Matsushita, Naoki Hayashi, Aiko Furukawa, Gaku Shoji, Takanobu Suzuki. "Trial of damage prediction for telecommunication conduits using machine learning". 17 World Conference on Earthquake Engineering (17 WCEE), Sendai, Miyagi, 2020, pp.1-10, 6a-0011.

International Conferences (not reviewed)

  1. Naoki Hayashi. "Bayesian Generalization Error and Real Log Canonical Threshold in Non-negative Matrix Factorization and Latent Dirichlet Allocation". Algebraic Statistics 2020, Zoom (virtual mini conference). June. 22 - 26, 2020. (2020/6/25). slideshare.

Conferences

  1. Naoki Hayashi. "The Exact Asymptotic Form of Generalization Error in LDA", IBIS2020, Online. (2020/11/26). in Japanese.
    • This talk is selected as one of Outstanding Presentation Finalists!!

Invited Talks

  1. Naoki Hayashi. "Theoretical Analysis of Bayesian Generalization Error in Parameter-restricted Matrix Factorization", StatsML Symposium'20, Online. (2020/12/5). in Japanese.

2019

Conferences

  1. Akira Saigo, Naoki Hayashi, Kotaro Ito. "Lesion Detection in Computed Tomography Images using YOLOv3 and Domain Knowledge", The 33rd Annual Conference of the Japanese Society for Artificial Intelligence 2019(JSAI2019), Niigata. (2019/6/5). in Japanese.

2018

Conferences

  1. Naoki Hayashi, Sumio Watanabe. "Experimental Analysis of Real Log Canonical Threshold in Stochastic Matrix Factorization using Hamiltonian Monte Carlo Method", Neurocomputing(NC), Tokyo. IEICE Technical Report, vol. 117, no. 508, NC2017-89, pp. 127-131. (2018/3/14). in Japanese.
  2. Naoki Hayashi. "Variational Approximation Accuracy in Non-negative Matrix Factorization", IBIS2018, Sapporo. IEICE Technical Report, vol. 118, no. 284, IBISML2018-51, pp. 53-60. (2018/11/5). in Japanese.

2017

Papers

  1. Naoki Hayashi, Sumio Watanabe. "Upper Bound of Bayesian Generalization Error in Non-Negative Matrix Factorization", Neurocomputing, Volume 266C, 29 November 2017, pp.21-28. doi: 10.1016/j.neucom.2017.04.068. (2016/12/13 submitted. 2017/8/7 published on web).

International Conferences

  1. Naoki Hayashi, Sumio Watanabe."Tighter Upper Bound of Real Log Canonical Threshold of Non-negative Matrix Factorization and its Application to Bayesian Inference". 2017 IEEE Symposium Series on Computational Intelligence (IEEE SSCI 2017), Honolulu, Hawaii, USA. Nov. 27 - Dec 1, 2017. pp.1-8, doi: 10.1109/SSCI.2017.8280811. (2017/11/28). Slides.slideshare.

Conferences

  1. Naoki Hayashi, Sumio Watanabe. "Experimental Analysis of Real Log Canonical Threshold in Non-negative Matrix Factorization", Neurocomputing(NC), IEICE Technical Report, Vol.116, No.521, pp.85-90. (2017/3/13). in Japanese.
  2. Naoki Hayashi, Fumito Nakamura. "Experimental Analysis of Variational Bayesian Method in Model Selection of Gaussian Mixture Model by Singular Bayesian Information Criterion", Information-Based Induction Sciences and Machine Learning (IBISML), IEICE Technical Report, Vol.117, No.211, pp.19-26. (2017/9/15). in Japanese.
  3. Naoki Hayashi, Sumio Watanabe. "Real Log Canonical Threshold of Stochastic Matrix Factorization and its Application to Bayesian Learning", IBIS2017, IEICE Technical Report, Vol.117, No.293, pp.23-30.(2017/11/9). in Japanese.

2016

Conferences

  1. Naoki Hayashi,Sumio Watanabe."A Real Log Canonical Threshold of Nonnegative Matrix Factorization and Its Application to Bayesian Learning", IBIS2016, IEICE Technical Report Vol.116, No.300, pp.215-220.(2016/11/17). in Japanese.

Preprints (including accepted paper)

arXiv Preprint

  1. Naoki Hayashi, Yoshihide Sawada. "Upper Bound of Bayesian Generalization Error in Partial Concept Bottleneck Model (CBM): Partial CBM outperforms naive CBM". arXiv: 2403.09206
  2. Naoki Hayashi, Yoshihide Sawada. "Bayesian Generalization Error in Linear Neural Networks with Concept Bottleneck Structure and Multitask Formulation". arXiv: 2303.09154
  3. Naoki Hayashi. "The Exact Asymptotic Form of Bayesian Generalization Error in Latent Dirichlet Allocation". arXiv: 2008.01304
  4. Naoki Hayashi. "Variational Approximation Error in Bayesian Non-negative Matrix Factorization". arXiv: 1809.02963
  5. Naoki Hayashi, Sumio Watanabe. "Asymptotic Bayesian Generalization Error in Latent Dirichlet Allocation and Stochastic Matrix Factorization". arXiv: 1709.04212
  6. Naoki Hayashi, Sumio Watanabe. "Upper Bound of Bayesian Generalization Error in Non-negative Matrix Factorization". arXiv: 1612.04112

bioRxiv preprint

  1. Keita Harada, Naoki Hayashi, Katsushi Kagaya. "Individual behavioral type captured by a Bayesian model comparison of cap making by sponge crabs". doi: 10.1101/330787

medRxiv preprint

  1. Akira Endo, Mitsuo Uchida, Naoki Hayashi, Yang Liu, Katherine E. Atkins, Adam J. Kucharski, Sebastian Funk. "Within and between classroom transmission patterns of seasonal influenza among primary school students in Matsumoto city, Japan". doi: 10.1101/2021.07.08.21259917

OTHER PUBLICATIONS

2021

Doctor Dissertation

  1. Naoki Hayashi. "Statistical Learning Theory of Parameter-Restricted Singular Models". Tokyo Institute of Technology. 2021FY. (2021/4/22 final defence，2021/6/30 completion)．Japanese presentation (public defence) slide is here. The dissertation (in English) is here.

2018

Master Thesis

  1. Naoki Hayashi．"Asymptotic Bayesian Generalization Error in Non-negative Matrix Factorization"．Tokyo Institute of Technology．2017fy．(2018/1/10 submit，2018/2/1 defence, in Japanese)．Slide is here．