A research group consisting of Professor Kenichi Ohki (affiliated with Institute for AI and Beyond (Beyond AI), Graduate School of Medicine, and International Research Center for Neurointelligence (WPI-IRCN), The University of Tokyo), Professor Teppei Matsui (Graduate School of Brain Science, Doshisha University (at the time of this research: Lecturer, Graduate School of Medicine, University of Tokyo)), Assistant Professor Takayuki Hashimoto, and Assistant Professor Tomonari Murakami (Graduate School of Medicine, The University of Tokyo), and Assistant Professor Masato Uemura (Faculty of Medicine, Kansai Medical University (at the time of this research: Project Assistant Professor, Graduate School of Medicine, The University of Tokyo)) have discovered a new mechanism by which the cerebral visual cortex separates visual information from spontaneous activity.

It is known that the brains of organisms, including humans, exhibit spontaneous activity generated by the brain itself even in the absence of sensory input (e.g., during rest). Spontaneous activity has been observed in various animals and is considered one of the most characteristic features of the biological brain. If spontaneous activity cannot be distinguished from activity in response to sensory input, it could lead to incorrect perceptions or hallucinations. However, in reality, the brain of living organisms can accurately process sensory information. The mechanism behind this has remained largely unexplained. Particularly, it was unclear how spontaneous activity is processed in the visual cortex of mammals, which forms a hierarchical network of multiple areas. Therefore, this study used marmosets, small primates with a visual system closer to that of humans and capable of advanced neural activity measurements, to analyze in detail the relationship between spontaneous activity and visual responses in multiple visual-related areas of the cerebral cortex. The research group discovered that within the hierarchical network of the cerebral visual cortex, the patterns of spontaneous activity and visual responses are similar in lower hierarchical levels, but gradually separate as one moves from lower to higher hierarchical levels. This gradual separation along the hierarchy suggests that the hierarchical nature of the network plays an important role in the separation of spontaneous activity and visual responses. This study not only proposes a new information processing mechanism of the cerebral cortex but also has the potential to contribute to the development of artificial intelligence that incorporates the advantages of the biological brain.

This research was supported by the Beyond AI Joint Project which is a collaborative initiative launched by the University of Tokyo and SoftBank, Japan Agency for Medical Research and Development (AMED), Japan Science and Technology Agency (JST), and Grants-in-Aid for Scientific Research (JSPS）.

The findings of this research was published in the British scientific journal "Nature Communications" on December 4 2024 (UK time).

・The full press release (in Japanese only) in PDF.
https://www.t.u-tokyo.ac.jp/hubfs/press-release/2024/0207/003/text.pdf

＜RELATED LINKS＞
・nature com.
https://www.nature.com/articles/s41467-024-54322-x

・UTokyo Focus
https://www.u-tokyo.ac.jp/focus/ja/press/

To the University of Tokyo faculty,

Beyond AI Joint Project with SoftBank aims to build an ecosystem for the social implementation and commercialization of research results through the Institute for AI and Beyond.
Currently, the Institute promotes mid-to-long-term (basic research) and “high-cycle” research (applied research). In high-cycle research, we aim to achieve social implementation and commercialization through the establishment of collaborative innovation partnerships (CIP) and eventually the establishment of companies, contributing to the further development of AI research in Japan and the realization of a better society.

The Institute for AI and Beyond has decided to call for proposals from research groups (individuals are also acceptable, but not students only) to solve issues faced by SoftBank or propose new research ideas to promote the industrialization of themes related to generative AI.
The following will be provided free of charge by SoftBank for conducting research:

1. Large-scale AI computing "NVIDIA DGX SuperPOD" (free of charge)
2. Support for building the necessary environment to utilize the above SuperPOD (free of charge)
3. Creation of AI learning data necessary for research (free of charge only if requested to be created by Softbank)

In addition, research budgets will be provided within the framework of pre-high-cycle research in Beyond AI Joint Project with SoftBank.

Overview of the Call for Proposals (Pre-High-Cycle Research)
■ Research Period: In principle, within one year (if crossing fiscal years, settlement is done annually)
■ Research Budget (excluding the above free provisions): Direct costs up to 10 million yen per project
■ Number of Projects to be Promoted: Approximately 6 projects
■ Eligibility: Faculty members (including project faculty members) and researchers (such as project researchers) with an employment relationship with the University of Tokyo
■ Application Deadline: By 18:00 on Monday, January 20, 2025 (JST)
■ Others: Decision on acceptance or rejection will be made after interview by the institute and consultation with the SoftBank researchers.
Themes other than those being recruited are also accepted, but themes that cannot create synergy with SoftBank may not be adopted.

※ Please be sure to check the following PDF (jn Japanese) for details (including how to apply).
https://beyondai.jp/contents/wp-content/uploads/2024/11/call_for_superpoduser.pdf

Please send any queries you may have regarding the invitation for proposals to the following address:
bai.kenkyushien.adm[at]gs.mail.u-tokyo.ac.jp
Please change [at] to @
(Research Support Team at the Institute for AI and Beyond)

＜RELATED LINKS＞
https://www.softbank.jp/corp/news/press/sbkk/2023/20231031_01/
https://www.nvidia.com/ja-jp/data-center/dgx-superpod/

A research group consisting of Associate Professor Takahiko Makiuchi, Assistant Professors Tomosato Hioki and Yuki Shimizu, Graduate Student Koujiro Hoshi, and Professor Eiji Saitoh (also affiliated with IMR (Institute for Materials Research, Tohoku University）, as Chief Researcher, and concurrently serving as Professor at the Institute for AI and Beyond, The University of Tokyo, collaborated with Mehrdad Elyasi, Assistant Professor at IMR (Institute for Materials Research, Tohoku University） (WPI-AIMR), and Gerrit Ernst-Wilhelm Bauer, Chief Researcher, to discover a mechanism that allows for persistent magnetic coherence in magnets, which was conventionally believed to last only for a short time.

The coherence within magnets has the potential to handle both 0 and 1 information, but maintaining this state for an extended period was considered challenging, limiting its practical applications. However, by successfully extracting the hidden coherence within magnets in this study, new possibilities have opened up for the development of magnetic information devices.

This research was supported by the Beyond AI Joint Project which is a collaborative initiative launched by the University of Tokyo and SoftBank, CREST-JST, ERATO-JST and Grants-in-Aid for Scientific Research (JSPS）.

The findings of this research was published in the British scientific journal "Nature Materials" on February 6, 2024 (UK time).

・The full press release (in Japanese only) in PDF.
https://www.t.u-tokyo.ac.jp/hubfs/press-release/2024/0207/003/text.pdf

＜RELATED LINKS＞
・nature com.
https://www.nature.com/articles/s41563-024-01798-z

・UTokyo Focus
https://www.u-tokyo.ac.jp/focus/en/press/

A research team consisting of Professor Kenichi OHKI and Jumpei UKITA (a graduate student at the time of the research) of the Department of Physiology, Division of Functional Biology, Graduate School of Medicine, the University of Tokyo, revealed that some of the vulnerabilities in deep neural networks can be mitigated by injecting random noise that mimics the brain in deep neural networks.

Currently, artificial intelligence (AI) is evolving at an accelerating pace, and its underlying structure is based on deep neural networks. Deep neural networks are known to be tricked into producing outputs that are clearly different from that of humans by malicious attacks, known as adversarial attacks. For example, image recognition AI in self-driving cars must correctly recognize a "stop" road sign as "stop" to make the car stop. However, the image recognition AI may not be able to correctly recognize a "stop" road sign generated by a hostile attack, even if it is clearly a "stop" sign when seen by a human. As a result, the car may not be able to stop, leading to a traffic accident. Thus, vulnerability to hostile attacks is one of the major challenges of implementing AI in society.

Incorporating brain properties of animals, such as humans, into AI may help overcome such vulnerabilities. The research team discovered that certain types of vulnerability can be reduced by introducing noise that mimics the randomness of neurons in the brain into deep neural networks. By using this method, the possibility of creating AI that more closely resembles the behavior of humans and other animals is thought to be greater.

This research was supported by the Institute for AI and Beyond of the University of Tokyo, the Japan Agency for Medical Research and Development (AMED) "Project to Elucidate the Entire Functional Brain Network through Innovative Technology", the Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, and CREST-JST.
The research result was published in Neural Networks on 16 September 2023 online.

Professor Ohki is the Project Leader for the Basic Research Project entitled “Development of next generation AI by modeling brain information” at the Institute for AI and Beyond.

・The full press release (in Japanese only) in PDF.
https://beyondai.jp/contents/wp-content/uploads/2023/09/Press-Release_NeuralNetworks_20230916_Web.pdf

＜RELATED LINKS＞
・Published paper (Title: Adversarial attacks and defenses using feature-space stochasticity)
https://www.sciencedirect.com/science/article/pii/S0893608023004422?via%3Dihub
DOI 10.1016/j.neunet.2023.08.022

・Article in English at UTokyo Focus
https://www.u-tokyo.ac.jp/focus/en/press/z0508_00311.html

The 2nd GENEE Symposium titled “Social Challenges Brought by Generative AI: Data Bias, Fake News, and Disordered Legal Systems” was held on August 3, 2023. The event was hosted by Generative AI and Educational Environment Research Project (GENEE) of the Collaborative Research Organization of Educational Technology at the University of Tokyo and co-hosted by the B’AI Global Forum.
Profs. Kaori Hayashi, Tohko Tanaka, and Yuko Itatsu of the B'AI Global Forum were among the speakers at the symposium.

For further details ;
https://baiforum.jp/en/events/en048/

A press release was issued on August 3, 2023 regarding new findings on a study on autism spectrum disorder (ASD) using the ASD perceptual experience simulator, in which Project Professor Yukie Nagai of the International Research Center for Neurointelligence (IRCN) at the University of Tokyo was involved as the principle investigator.

Professor Nagai was the former leader of the Mid- and Long-term Research Project “AI x Tojisha-Kenkyu: Computational Neuroscience for Systematic Understanding of Cognitive Individuality” and current Alumni committee member of the Institute for AI and Beyond.

Full press release at the IRCN official website:
https://ircn.jp/pressrelease/2023083_yukie_nagai

Professor Nagai presented the ASD perception experience simulator in the 8th Beyond AI Science Café video, in which she was featured last year.
The video, which is available in Japanese at the Institute for AI and Beyond YouTube Channel as linked below;

A scientific paper entitled “Audio Signal Enhancement with Learning from Positive and Unlabeled Data” written by Dr. Nobutaka ITO and Professor Masashi SUGIYAMA of the Graduate School of Frontier Science at the University of Tokyo was selected for the “Best Paper Award” at 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2023), one of the largest and most influential research conferences of the Signal Processing Society in the world. The paper was selected from 6127 submitted papers, of which 2765 were accepted for publication in the conference proceedings.

Dr. Ito who participated in-person in the conference, received the award at the awards ceremony held on Friday, June 9, in Rhodes, Greece from the organizing committee, represented by General Chair Dr. Petros Maragos.

Professor Sugiyama and Dr. Ito are respectively, the project sub-leader and associate researcher of the Basic research project (Advancement of Fundamental AI) titled “Automatic Learning of High Accurate Prediction Models from Limited Supervised Data” led by Professor Tatsuya HARADA (Research Center for Advanced Science and Technology, The University of Tokyo) in the Institute for AI and Beyond.

Dr. Ito receiving the honorable certificate from Petros Maragos General Chair

・Full paper
https://ieeexplore.ieee.org/document/10095988/authors#authors
・ICASSP2023 Official website
https://2023.ieeeicassp.org/

An international team of astrophysicists and cosmologists from various institutes, including Professor Masahiro Takada and Dr. Sunao Sugiyama, a Project Researcher, of the Kavli Institute for the Physics and Mathematics of the Universe (Kavli IPMU) at the University of Tokyo, submitted a set of five papers that suggest the possibility that the standard cosmological model proposed by Einstein may be incomplete in some interesting way.

Professor Takada commented on the application of AI to the experiment as follows:
“In this study, we used interim data from the Subaru Telescope Hyper Suprime-Cam survey to measure a cosmological phenomenon called weak gravitational lensing effect. We then compared the measurement results with the cosmological model to measure physical quantities that describe the universe, called cosmological parameters.
In this theoretical model, we constructed a regression model using deep learning, with a large amount of data from cosmological simulations as training data. In other words, the theoretical model is equivalent to a universe created on a supercomputer and is equivalent to comparing Subaru’s observations with a supercomputer’s theory. Using this, we measured the cosmological parameters and found discrepancies with the data of the early universe.”

Professor Takada and Dr. Sugiyama are the Sub-Leader and Associate Researcher, respectively, of the Basic research project “New developments in AI through study of evolution of the Milky Way from initial quantum fluctuations to its assembly” (Research Leader: Professor Hitoshi Murayama of Kavli IPMU) at the Institute for AI and Beyond.

Full press release at the Kavli IPMU Official Website:
https://www.ipmu.jp/en/20230404-darkmatter
Project page of Professor Murayama's Research Group at the Institute for AI and Beyond:
https://beyondai.jp/contents/projects/murayama/?lang=en

Follow @BeyondAI_UTokyo

Professor Eiji Saitoh of the Graduate School of Engineering, leader of the Mid- and Long-term research project "Analysis of materials’ quantum properties using AI (Quantum ID: Exploiting “Quantum Fingerprintings” of materials with AI)", has posted a column on his website regarding his research at the Institute for AI and Beyond.
The column is titled “New Challenges with Machine Learning - Are AI and Physics a Good Match?” and written by Marie Minagawa, his research collaborator.

・Link to the column (available in Japanese only):
https://saitoh.t.u-tokyo.ac.jp/interview/column/entry-892.html
・Prof. Saitoh’s Research Project page:
https://beyondai.jp/contents/projects/saitoh/?lang=en

HEMILLIONS Inc., a commercial company, was established on 7th April as a joint initiative of the University of Tokyo, SoftBank Corp., and KIS Inc. The company will promote health and medical data standardization and their dissemination by utilizing the HL7 FHIR (FHIR: Fast Healthcare Interoperability Resources) standard. Ultimately, HEMILLIONS will work towards building a society that can utilize health and medical data safely and securely by developing and providing solutions such as FRUCtoS, a platform software including FHIR server.

The HEMILLIONS started as an applied research project, a part of the efforts undertaken by the Institute for AI and Beyond, and was later converted to a joint venture company through the Collaborative Innovation Partnership (CIP) system. It is the first commercialized and incorporated company initiated by the Institute. It is also the first case for the University of Tokyo to directly hold common shares of a company in return for their intellectual properties.

The Institute of AI and Beyond aims to create new academic fields by integrating fundamental technology research of AI and other research fields. It also aims to establish an ecosystem which can return resources to the University to enhance its education and research activities by accelerating the social implementation and business deployment of the research outcomes generated from the Institute. The Institute will continue its efforts in commercialization based on the research outcomes via the CIP system to further promote the ecosystem.

The full press release is available below (in Japanese only) (PDF).
https://beyondai.jp/contents/wp-content/uploads/2023/04/20230420_Press-Release_HEMILLIONS.pdf

UTokyo Focus　https://www.u-tokyo.ac.jp/focus/ja/press/z1702_00042.html
SoftBank News https://www.softbank.jp/corp/news/press/sbkk/2023/20230420_01/