Yato Shoma

About

I am a fourth-year undergraduate student working on deep learning and generative models. My current research topics include:

• Undergraduate Thesis: Can diffusion-model-based generative augmentation of minority-class patches (e.g., positive / rare tissue types) in WSI statistically improve slide-level MIL classification metrics (PR-AUC, Macro-F1, Balanced Accuracy)?
  Official Japanese title: 「WSIの少数クラス（陽性・希少組織型など）パッチの拡散モデル生成拡張による、スライドレベルのMIL分類指標（PR-AUC, Macro-F1, Balanced Acc）の統計的改善が可能か」
• Investigating memorization and forgetting in LLM pretraining (as a Technical Assistant at the National Institute of Informatics, LLM Center)

I enjoy AI / machine learning as well as creative work, and I am also an operating member of the student community “AcademiX.”

Learning & Activities

Affiliations / Work Experience

• Tokyo Polytechnic University, Faculty of Engineering, Department of Engineering (Apr 2022–present)
• Matsuo Lab LLM Community “LLMATCH” (Cohort 2), Community Researcher (Apr 1–Sep 30, 2025)
• National Institute of Informatics (NII), LLM Center — Technical Assistant (Dec 2025–present) （技術補佐員）

Completed Programs

Matsuo–Iwasawa Laboratory (The University of Tokyo)

• GCI 2023 Winter — Completed
• Deep Learning Basic 2024 — Completed
• Large Language Models (LLM) 2024 — Completed
• Deep Generative Models 2024 — Completed
• World Models 2024 — Completed (Outstanding Final Project Presentation) （世界モデル2024）
• AI Management Program 2025 — Completed
• Deep Learning Basic 2025 — Completed
• Summer School 2025: Deep Reinforcement Learning — Completed
• Summer School 2025: Deep Generative Models — Completed
• AI and Semiconductors 2025 — Completed (Top Performer in Exercises) （演習優秀生）

Open Courses

• AI Training Program (AcademiX) — Completed
• Stanford CS231n Study Group (two rounds)
• MIT 6.5940 TinyML and Efficient Deep Learning

Programs in Progress

Matsuo–Iwasawa Laboratory (The University of Tokyo)

• Large Language Models (LLM) 2025 — Fundamentals (In progress)
• World Models 2025 (In progress)
• Deep Learning Fundamentals 2025 Autumn (In progress)

Open Courses in Progress

• CS492(D): Diffusion Models and Their Applications

Research

2026

• Official Japanese title: 「官庁出版物コーパスを用いた日本語 LLM の継続事前学習とその分析」
  English translation: “Continual Pretraining and Analysis of a Japanese LLM Using a Government Publications Corpus”
  屋藤翔麻 (東京工芸大), 清丸寛一, 小田悠介 (NII), 河原大輔 (早稲田大学)
  The 267th Natural Language Processing Research Meeting (IPSJ SIG-NL), March 7, 2026, Program
  Japanese venue name: 第267回自然言語処理研究発表会

2025

• “Improving Exploration Efficiency in Sparse-Reward Environments via Refined Intrinsic Motivation and Contrastive Learning”
  ┗ Conducted as the final project for the World Models / Deep Learning Applications Course (Matsuo–Iwasawa Laboratory, UTokyo).
  ┗ Presented at the following conferences:
  • Official Japanese title: 「内発的動機付けと対照学習の改良によるスパースな報酬環境における探索効率向上」
    English translation: “Improving Exploration Efficiency in Sparse-Reward Environments via Refined Intrinsic Motivation and Contrastive Learning”
    Hiroki Sagara, Shoma Yato, Shuichi Kusaba
    The 39th Annual Conference of the Japanese Society for Artificial Intelligence (JSAI 2025), Osaka International Convention Center + Online, May 27–30, 2025, Session ID: 2Win5-14, DOI: 10.11517/pjsai.JSAI2025.0_2Win514
  • Improving Exploration Efficiency in Sparse-Reward Environments via Refined Intrinsic Motivation and Contrastive Learning,
    Hiroki Sagara, Shoma Yato, Shuichi Kusaba, Yousun Kang,
    The 40th International Technical Conference on Circuits/Systems, Computers and Communications (ITC-CSCC 2025), July 8, 2025
• Matsuo Lab LLM Community “LLMATCH” Cohort 2: “Structured Memory Enhancement for Generative Agents Using Knowledge Graphs”
  Slides Recording

AI Development

2024

• The University of Tokyo May Festival 2024 — Virtual Cafe AI App Development （五月祭）

2025

• Voice Imitation AI App: VALL-E-X_JP-Voice-Cloner
• A lightweight ONNX-based toolkit: Waifu-GAN-Flow
• An AI That Talks! Othello Commentary AITuber System — Tokyo Polytechnic University Festival 2025 Exhibition Project （工芸祭2025）

2026

• JSAI 2026 40th Anniversary Special Session 6: AIWolf Contest 2026 Domestic Tournament
  Japanese official name: 40周年記念企画6：人狼知能コンテスト2026国内大会
  2026 Spring Domestic Tournament, Natural Language Processing Division
  Japanese official division name: 2026春季 国内大会 自然言語処理部門
  yatolab: “Improving an AIWolf Agent through Character Assignment and Separated Self/Other Models”
  Japanese project title: yatolab: キャラクター性付与と自己・他者モデル分離による人狼知能エージェントの改良
  Akika Itakura, Ikumu Komatsu, Shoma Yato (Tokyo Polytechnic University) （板倉明佳，小松生夢，屋藤翔麻（東京工芸大学））

Hackathons

2024

• What-If Story Generator – A “What if…?” prompt → storytelling assistant powered by RAG and Gemini 1.5 Pro, outputs in Markdown format 🌟

2025

• Tokyo AI Festival 2025 Hackathon
  • Disaster Support Web App Using a Local LLM

Projects & Works

Works / Video Credits

• Atsugi City Administrative Content (2022) （厚木市行政コンテンツ2022）
  
  • “Reunion in Atsugi — Atsugi OEC Food” (Filming & Editing) （「厚木で再会～あつぎＯＥＣフード～」）
  • “Architectural History of Atsugi — The History of Atsugi” (Composition, Editing, Color Grading) （「厚木の建築史～厚木の歴史～」）
• Cooperation for the 70th Anniversary Project of Atsugi City Administration (Colorization of black-and-white photos) （「厚木市政70周年記念事業協力」）

Awards

• Excellence Award for a submitted report in “Social Studies by Students” （「学生による社会スタディ」優秀証）

Hobbies & Skills

• AI Development: CNN, deep learning model implementation, web app development (React, Vue.js)
• Creative Work: illustration, music production (Hatsune Miku), 3DCG, DJ, etc.
• Tools: Anaconda, Docker, AWS, team development with GitHub

Links

Contact

Email: e2213162@st.t-kougei.ac.jp

SNS / Blog