Satoki Ishikawa / 石川智貴

About

• School of Computing, Tokyo Institute of Technology
• Lab: Yokota Lab [link]

Research

• Deep Learning
• Second-order Optimization
• Large-scale Distributed Parallelization

Publications

Reviewed (Conference)

• Satoki Ishikawa, Rio Yokota, Ryo Karakida. Local Loss Optimization in the Infinite Width: Stable Parameterization of Predictive Coding Networks and Target Propagation. ICLR2025
  
  • arxiv [link]
• Satoki Ishikawa, Makoto Yamada, Han Bao, Yuki Takezawa. PhiNets: Brain-inspired Non-contrastive Learning Based on Temporal Prediction Hypothesis. ICLR2025
  
  • arxiv [link]
• Satoki Ishikawa, Ryo Karakida. On the Parameterization of Second-Order Optimization Effective Towards the Infinite Width. ICLR2024
  
  • arxiv [link]

Reviewed (Workshop)

• Satoki Ishikawa, Tal Ben-Nun, Brian Van Essen, Rio Yokota, Nikoli Dryden. Lion Cub: Minimizing Communication Overhead in Distributed Lion. ICML2025 Tiny Titans: The next wave of On-Device Learning for Foundation Models
  
  • arxiv [link]
• Satoki Ishikawa, Rio Yokota. When Does Second-Order Optimization Speed Up Training?. ICLR2024 TinyPaper Track
  
  • openreview [link]
• Kazuki Osawa, Satoki Ishikawa, Rio Yokota, Shigang Li, Torsten Hoefler. ASDL: A Unified Interface for Gradient Preconditioning in PyTorch. HOOML2022.
  
  • HOOML2022 [link]
  • arxiv [link]

Domestic Conferences

• Satoki Ishikawa, Rio Yokota. "Study on Gradient Preconditioning Methods in Deep Learning." IPSJ 85th Annual Conference 2023 [Student Encouragement Award]
• Paper Link [link]

Education

• 2023.04 - Current
  
  • Master's Program, School of Computing, Tokyo Institute of Technology
• 2019.04 - 2023.03
  
  • Bachelor's Program, School of Computing, Tokyo Institute of Technology
  • Graduation Representative

Awards

• Student Encouragement Award, IPSJ 85th Annual Conference
• Excellent Student Award, School of Computing, Tokyo Institute of Technology

Links

• Twitter
  
• GitHub
  

Contact

• ishikawasatoki [at mark] outlook.jp