Naoto Iwase
岩瀬 直人
Naoto Iwase is a medical student at Nagoya University and a part-time engineer at Preferred Networks, Inc. His research interests lie in machine learning methods, with a particular focus on large language models and their clinical applications.
He also writes ML Notes, a collection of survey-style notes on recent machine learning papers.
Research Interests
- Reliable LLM reasoning: chain-of-thought verification and statistical inference for self-consistency
- Medical AI benchmarks and clinical natural language processing
- Computational pathology: cell-niche prediction and multi-task learning on H&E images
First-author Works
-
Reliable Chain-of-Thought via Prefix Consistency
Correct chain-of-thought traces reproduce their answer under prefix regeneration more often than wrong ones; weighting majority voting by this prefix consistency reaches plateau accuracy at up to 21× fewer tokens (median 4.6×).
[paper] [project page] [code]
-
MedRECT: A Medical Reasoning Benchmark for Error Correction in Clinical Texts
A bilingual (Japanese/English) benchmark for medical error correction built from licensing exams; across 9 LLMs, reasoning models substantially outperform standard architectures, and a fine-tuned model exceeds human expert performance.
Presentations
Awards
- Top Performance Award, Japan Statistical Society Certificate Pre-1st Grade, January 2025.
- Student Presentation Award (Oral), 38th Annual Meeting of the Japanese Society of Computational Statistics, May 2024.
- Top Score, Entrance Examination, Nagoya University School of Medicine (2021).