I am a researcher at Preferred Networks, Inc. (PFN) from July 2018. Before joining PFN, I had been working as a research scientist in machine learning at NTT Communication Science Laboratories for 10 years. Then I worked at Mirai Translate, a start-up for machine translation business, from July 2016 to June 2018, at the same time served as a project manager at NTT DoCoMo.
Short summary and News
- Started my career at NTT Commnication Science Labs. as a scientific researcher in statistical machine learning.
- In Mirai Translate, I was the leader for all rsearch issues concerning the performance of the Neural Machine Translation (NMT) engine, which was (possibly) the first NMT engine product developed, trained, tuned, integrated, sold, and maintained by a Japanese private company.
- Now I am working as a researcher in PFN, joining several exciting but diverse projects. Especially, I have been working on Graph Neural Networks (GNNs).
- My book is out now! "関係データ学習" (Learning from relational data) published from Kodansha Scientific, Japan. See more in Amazon .
- My most recent paper is published now in arXiv, ``Weisfeiler-Lehman Embedding for Molecular Graph Neural Networks'', with Dr. Oono and Dr. Hayashi (Preferred Networks). A handy graph preprocessors for molecular graphs. Theoretically assured better trainability and experimentally validated generalization performance, applicable to many GNNs.
- Our paper, ``Learning Structured Latent Factors from Dependent Data:A Generative Model Framework from Information-Theoretic Perspective'' has been accpted at ICML 2020! This work generalizes several structured latent variable models e.g. beta-VAE, CorEX, and their variants under the unified view from Information Bottleneck.
- One of my tutorial talks of GNNs in English is available at SlideShare. This is a lecture material for: a special lecture on Data Science at Nara Institute of Science and Technology (NAIST).
Selected Publicationss (See the publication page for available links)
Weisfeiler-Lehman Embedding for Molecular Graph Neural Networks,
arXiv: 2006.06909 [cs.LG], 2020.
GraphNVP: An Intertible Flow Model for Generating Molecular Graphs,
arXiv: 1905.11600 [cs.LG], 2019.
Learning Structured Latent Factors from Dependent Data:A Generative Model Framework from Information-Theoretic Perspective,
Proceedings of the 37th International Conference on Machine Learning (ICML2020), 2020.
Averaged Collapsed Variatonal Bayes Inference,
Journal of Machine Learning Research, Vol. 18. No.1, pp. 1-29, 2017.
Subset Infinite Relational Models,
Proceedings of the 15th International Conference on Artificial Intelligence and Statistics (AISTATS2012), pp. 547-555, 2012.
Dynamic Infinite Relational Model for Time-varying Relational Data Analysis,
Advances in Nueral Information Processing Systems 23 (Proceedings of NIPS2010), 2010.
Resources and Contacts
- Curriculum Viate: Available HERE (last update: May, 2021)
- My Links on Github, Google Scholar, Linkedin, Hatena Blog (mostly Japanese posts)
- E-Mail: k 'dot' ishiguro 'dot' jp 'AT' ieee 'dot' org
- Postal adress: Preferred Networks, Inc. Otemachi Bldg. 3rd floor, 1-6-1 Otemachi, Chiyoda-ku, Tokyo 100-0004 Japan