September 06, 2022 Notices

External Research/Presentation Reports(22’ July-August)

ARAYA R&D department members gave presentations and gave lectures during July and August. Please take a moment to learn more about the topics of your interest.

Manuel Baltieri, Ph.D., gave a presentation.
Title: “The Emperor’s New Markov Blanket”
Date:13th July 2022
For more info:
Web site

Consciuosness Club Tokyo
Manuel Baltieri, Ph.D., lead a special session at the International Conference on Artificial Life (ALIFE 2022)
Date:8th July 2022
Title: “Inference with and within a model”
For more info:
Web site

ALIFE 2022
Manuel Baltieri, Ph.D., lead a special session at the International Conference on Artificial Life (ALIFE 2022)

Title: “Hybrid life V: Approaches to integrate biological, artificial and cognitive systems”
Organizers:​ Manuel Baltieri, Keisuke Suzuki and Olaf Witkowski
Date: 20th July 2022
The session focuses on hybrid methods studying artificial, living and cognitive systems, looking in particular at 1) theoretical contributions across fields (agency, adaptation, etc.), 2) systems augmentation and enhancement, and 3) interactive heterogeneous systems.
For more info:

Invited keynote presentation:
Title: "Robotic human movement augmentation
Speaker: Yanpei Huang and Jonathan Eden
For more info:

Tohoku University Research DX Strategy Seminar No. 4
Hiro Hamada, Ph.D., was invited for a Seminar.
Title: The Future of Open Science Pioneered by Blockchain Technology - Incentive Design through Decentralized Science
Date: August 2, 2022
For more info
Web page:
Presentation slides:

Hokkaido University Center for Human Nature, Artificial Intelligence, and Neuroscience. CHAIN ACADEMIC SEMINAR
Ippei Fujisawa, Ph.D., was invited for a Seminar.
Title: Logic tasks for extrapolation ability and rule understanding
Date: August 23, 2022
Abstract :
Logical reasoning is essential in a variety of human activities. Mathematics is often regarded as a typical logical task. Although large language models such as GPT-3 have succeeded in various fields, their accuracy for elementary arithmetic tasks is limited. Since humans understand arithmetic tasks, it is easy to construct systems which solve specific tasks by implementing our knowledge. In this talk, we will review these situations and discuss our proposal for characterizing a broader class of logic tasks, including arithmetic tasks. We will also discuss relations to extrapolation and explainability. We expect that logic tasks will help us to find suitable inductive biases for logical processing.
 For more info: