雑記 ー Fumito Uwano's pages ー

My students, Mr. Murakami and Mr. Matsumoto, presented their research at the Japanese conference SSI 2025.
Mr. Murakami received the SSI Research Encouragement Award.

SSI (SICE Systems and Information Division Annual Conference) is the annual conference of the Systems and Information Division of the Society of Instrument and Control Engineers (SICE). It is one of the major domestic conferences in system engineering and informatics in Japan.

In addition to their participation in DEIM 2025, they delivered excellent presentations at SSI 2025 and demonstrated strong research capabilities, which led to this recognition. I am very pleased to support their academic growth.

Conference Details

• Conference name: SICE Systems and Information Division Annual Conference (SSI 2025)
• Location: Tohoku University, Aoba Ward, Sendai, Japan
• Date: November 12th to 14th, 2025

Presentation Details

• Title: LLM-Supported Reinforcement Learning in Sparse Reward Environments
• Authors: Ryotaro Murakami, Manabu Ohta, and Fumito Uwano
• Title: Multi-Vessel Path Finding Method Using Reinforcement Learning Toward Real-World Environments
• Authors: Tenyu Matsumoto, Manabu Ohta, and Fumito Uwano

Conference Atmosphere (Left: Presentation, Right: Award Ceremony)

I gave an oral presentation in ICAART 2022. This conference deals with specialties of agent technology.

Agent technology can cover large field where human-agent interaction, robotics, data mining, etc.
The invited speaker who has a specialty for human-agent interaction gave us an interesting presentation. As for the resent work, she introduced agents interact to a human for any senses in VR. I reckoned it’s marvelous in pandemic.
It was a first time for me to give a presentation. There is an atmosphere for the international conference because almost all of the presenters come from several countries out of Japan.

I told the participants the proposal how agents with difference resolution inputs learn to cooperate each other and the experimental results. I had seven minutes for presentation, which I thought there is no question from my presentation. However, participants would ask me the details of the problem domain.
Although it was hard for Japanese to hear all presentations in GMT time zone, I will hear the remain presentations to improve my research.

Presentation Information

• Title: LSTM-based Abstraction of Hetero Observation and Transition in Non-Communicative Multi-Agent Reinforcement Learning
• Author: Fumito Uwano

Conference Information

• Conference Name: 14th International Conference on Agents and Artificial Intelligence (ICAART 2022)
• Location: Zoom (online)
• Date: February 4th to 6th, 2022

I gave an oral presentation in AROB2022. This symposium deals with specialties of artificial life and robotics.

It was a first time I joined in. I had a great time as I had presentations given by close researchers. Furthermore, My colleagues’ supervised students gave me and others their presentations, which made me realize I had already started a career for an assistant professor.

I told participants the results about the research supported by KAKENHI Grant (Research Activity Start-up). That grant will be finished two months later, but the results made many kinds of valuable challenge beyond my plan.
I might have an opportunity to submit an application for advanced research from this theme to other grants, but I am going to finish this research as a milestone.

Presentation Information

• Title: Policy-oriented Goal Selection in Multi-Agent Reinforcement Learning for Dynamic Environments without Communication
• Author: Fumito Uwano

Conference Information

• Conference Name: 27th International Symposium on Artificial Life and Robotics (AROB 2022)
• Location: Zoom (online)
• Date: January 25th to 27th, 2022

I gave a poster presentation on Japanese conference named SSI2021. This conference is organized by SANDY in the Society of Instrument and Control Engineers (SICE).

I joined in this conference for many times. Since I can communicate and discuss with familiar professors in that conference, it is so valuable for my research topic.

By the way, my presentation explained about my research topic of KAKENHI grant for young researchers, and showed the previous methods’ performance on simple grid environments. That grant could support me until Mar. 2024, that is, next year, or the middle term of that grant, will get tough for me. Therefore I will do my research more deeply in next year.

Presentation Information

• Title: Examination of knowledge transfer for learning agents to adapt to unknown environment (in Japanese)
• Author: Fumito Uwano

Conference information

• Name: SSI 2021
• Place: oVice (Online)
• Date: November 20th to 22nd, 2021