Shota Saito

Ph.D student at University College London. Working on niche of graph machine learning.

I’m Shota, a Ph.D student at University College London. I work on graph machine learning. For the details of the profile see my website.

Education

  • 2019 - present, Ph.D Student, Department of Computer Science, University College London. Supervisors: Prof. Mark Herbster and Prof. John Shawe-Taylor.
  • 2012 - 2015, MSc, Department of Mathematical Informatics, The University of Tokyo
  • 2008 - 2012, BEng, Department of Mathematical Engineering and Information Physics, The University of Tokyo

Work Experience

  • Jan. 2021 - Mar. 2021, Research Intern at Microsoft Research
  • Sep. 2016 – Apr. 2019, Data Scientist at Amazon Web Services
  • Apr. 2015 - Aug. 2016, Software Eingeer at DeNA Co., Ltd.
  • Jan. 2014 - Mar. 2014, Research Intern at IBM Research Tokyo

Publication

See the full list at Google Scholar

Workshops/Preprints

  • Stephen Pasteris, Alberto Rumi, Maximilian Thiessen, Shota Saito, Atsushi Miyauchi, Fabio Vitale and Mark Herbster. Bandits with Abstention Under Expert Advice. 2024 [arXiv]

  • Shota Saito, Takanori Maehara, Mark Herbster. ResTran: A GNN Alternative to Learn A Graph with Features. In ICLR 2024 Workshop on Machine Learning for Genomics Explorations (MLGenX) [paper]

Conferences/Journals

  • Shota Saito and Mark Herbster. Multi-class Graph Clustering via Approximated Effective p-Resistance, In Proc. ICML, 29697-29733, 2023 [arXiv][code][poster][link][video]

  • Shota Saito and Mark Herbster. Generalizing p-Laplacian: Spectral Hypergraph Theory and a Partitioning Algorithm. Mach. Learn. (2023) [Link]

  • Shota Saito. Hypergraph Embedding: Hypergraph Cut, Kernel k-means, and Heat Kernel. In Proc. AAAI, 8141-8149, 2022 [arXiv][link]

  • Shota Saito, Danilo P Mandic, and Hideyuki Suzuki. Hypergraph p-Laplacian: A Differential Geometry View. In Proc. AAAI, 3984-3991, 2018 [arXiv][code]

  • Shota Saito, Yoshito Hirata, Kazutoshi Sasahara, and Hideyuki Suzuki. Tracking Time Evolution of Collective Attention Clusters in Twitter: Time Evolving Nonnegative Matrix Factorisation. In PLoS ONE 10(9): e0139085. [Link]

  • Shota Saito, Ryota Tomioka, and Kenji Yamanishi. Early Detection of Persistent Topics in Social Networks. Soc. Netw. Anal. Min. 5(19), pp.1-15, 2015 [pdf]

Workshops

  • Shota Saito, Yohei Ikawa, Hideyuki Suzuki and Akiko Murakami. Early Detection of Disasters with Contextual Information on Twitter. IEICE Tech. Rep., vol. 114, no. 81, NLC2014–2, pp. 7–12, 2014. (in Japanese) [pdf]