TAI AAI #04 - Brain-inspired AI
NOTE: registration on this page is required at most 24 hours before the event (to generate the entrance QR codes).
Topic
For decades, neuroscience research has inspired research in AI, and vice versa. Many researchers believe that neuroscience still has plenty to contribute to AI, and at Tokyo AI (TAI) we will host several researchers from academia and industry to present their work on brain-inspired AI, with topics ranging from AI agents implementing theories of consciousness, to developmental robotics.
Our Community
Tokyo AI (TAI) is a community composed of people based in Tokyo and working with, studying, or investing in AI. We are engineers, product managers, entrepreneurs, academics, and investors intending to build a strong “AI coreˮ in Tokyo. Find more in our overview: https://bit.ly/tai_overview
Schedule
17:30-18:00 Doors Open
18:00-18:30 Designing Embodied AI Agents based on Theories of Consciousness (Rousslan Dossa)
18:30-19:00 Cognitive Development Through Embodied Predictive Processing (Yukie Nagai)
19:00-19:45 Break/Networking
19:45-20:15 Brain Reference Architecture (BRA) Driven Approach and BRA Data Sharing (Yoshimasa Tawatsuji)
21:15-20:45 Proposal of BDBRA, a Literature Database to Accelerate BRA-Driven Development, and Efforts Toward Automating Database Construction (Yuta Ashihara)
21:00 Finish
Speakers
Rousslan Dossa (profile)
Title: Designing Embodied AI Agents based on Theories of Consciousness
Abstract: With their ever-increasing complexity, artificial neural networks have become able to approximate basic cognitive functions of the human brain. An evident progression would thus be to investigate how these networks can be structured to model higher-level cognitive processes such as consciousness. With this aim, we designed and empirically assessed an embodied agent with a structure based on indicator properties that underpin Global Workspace Theory, a leading theory of consciousness. Our agent was trained to navigate a 3D environment based on realistic audiovisual inputs. Among our findings, the agent with a Global Workspace architecture demonstrated greater robustness at smaller working memory sizes, as compared to agents with standard recurrent architectures. We also found task complexity and regularization as key components for the learning of meaningful features and attentional patterns within the workspace.
Bio: Rousslan is a Chief Researcher at Araya. He received his Ph.D. from Kobe University in 2023. His research interests span over the topics of deep reinforcement learning with an emphasis on self-supervised learning, human cognition-inspired decision-making, neuroscience, and evolutionary computing.
Yukie Nagai (profile)
Title: Cognitive Development Through Embodied Predictive Processing
Abstract: Cognitive development is an intricate and multifaceted process. Human abilities related to perception and action continually evolve during development, exhibiting remarkable diversity among individuals. This presentation explores the concept of embodied predictive processing as a promising unified theory for cognitive development. Rooted in neuroscience, predictive processing offers a unique perspective for understanding how the brain constructs its perception and actions with the environment. The core idea posits that the brain continually generates internal models to predict the world and refines them in response to sensory input to minimize prediction errors. This dynamic process underlies the acquisition of cognitive abilities, from self-recognition to goal-directed actions, and even fosters the emergence of social behaviors. Moreover, this presentation sheds light on how disruptions in predictive processing lead to individual diversities, including developmental disorders. By showcasing its practical application in robotic experiments, we aim to demonstrate potentials of embodied predictive processing as a unifying framework for cognitive development.
Bio: Yukie Nagai is a Project Professor at the International Research Center for Neurointelligence at the University of Tokyo. She earned her Ph.D. in Engineering from Osaka University in 2004, after which she worked at the National Institute of Information and Communications Technology, Bielefeld University, and then Osaka University. Since 2019, she has been leading the Cognitive Developmental Robotics Lab at the University of Tokyo. Her research encompasses cognitive developmental robotics, computational neuroscience, and assistive technologies for developmental disorders. Dr. Nagai employs computational methods to investigate the underlying neural mechanisms involved in social cognitive development. In acknowledgment of her work, she received the titles of "World's 50 Most Renowned Women in Robotics" in 2020 and "35 Women in Robotics Engineering and Science" in 2022, among other recognitions.
Yoshimasa Tawatsuji (profile)
Title: Brain Reference Architecture (BRA) Driven Approach and BRA Data Sharing
Abstract: Building brain-like artificial intelligence is a promising approach to engineering human-like problem-solving abilities. To this end, it is essential to understand and imitate the computational functions of the brain, which requires knowledge from a wide range of fields such as neuroscience and cognitive science. Based on Brain Reference Architecture (BRA)-driven development, we have constructed BRA data that organizes the structure and function of various brain regions. We also held the first international workshop aimed at disclosing, sharing, and integrating BRA data and data papers. In this presentation, we will provide an overview of our BRA-driven development and introduce our BRA data sharing efforts.
Bio: Yoshimasa Tawatsuji is an assistant professor at Graduate School of Engineering, the University of Tokyo. He received his Ph.D. in Human Sciences from Waseda University in 2020. He has worked as a research associate at the School of Human Sciences, Waseda University, an assistant professor at Global Education Center, Waseda University, and an assistant professor at Center for Data Science, Waseda University. His research interests include uncanny valley, model of others, and biologically plausible architecture for emotion.
Yuta Ashihara (profile)
Title: Proposal of BDBRA, a Literature Database to Accelerate BRA-Driven Development, and Efforts Toward Automating Database Construction
Abstract: WBAI has advocated BRA-driven development as a methodology for developing architectures inspired by the brain. In BRA-driven development, software development grounded in the brain can be advanced by using anatomical knowledge of the brain as a constraint. However, currently, it is necessary to collect anatomical information about the brain manually. Given the vast knowledge of the brain, collecting accurate information manually is considered challenging. To address this, we have been developing BDBRA, a database designed to extract anatomical knowledge of the brain from neuroscience literature automatically. In this talk, I will discuss how BDBRA was structured and introduce the future development roadmap for BRA-driven development, including BDBRA.
Bio: Yuta Ashihara is an assistant professor in the Department of Information Science at Nihon University, a professional academic staff member at the University of Tokyo's Graduate School, and a principal researcher at NGen Co., Ltd. He holds a master's degree in Informatics from the University of Electro-Communications. After founding Cross Compass Corporation and Glia Computing Corporation, he has worked as an engineer on projects applying deep learning for automation in the manufacturing and robotics industries. His area of expertise is machine learning.
Target Audience
Technical background with some knowledge of AI.
Organizers - alphabetic order
Kai Arulkumaran: Research Team Lead at Araya, working on brain-controlled robots as part of the JST Moonshot R&D program. Previously, he completed his PhD in Bioengineering at Imperial College London and had work experience at DeepMind, FAIR, Microsoft Research, Twitter Cortex, and NNAISENSE. His research areas are deep learning, reinforcement learning, evolutionary computation, and computational neuroscience.
Ilya Kulyatin: Fintech and AI entrepreneur with work and academic experience in the US, Netherlands, Singapore, UK, and Japan, with an MSc in Machine Learning from UCL.