AIP Acceleration Research
"AIP Acceleration Research" aims to maximize the research result as an AIP Network Laboratory by supporting newly-proposed research projects based on the excellent research results among the laboratory.
AIP Acceleration Research FY 2023
Construction of Artificial Olfactory Sensor System for Non-Invasive Detection and Early Treatment of Hard-to-Find Gynecological Diseases
Research Director
| Kazuki Nagashima |
Professor, Research Institute for Electronic Science, Hokkaido University |
Collaborators
| Takao Yasui |
Professor, Department of Life Sciences and Technology, Tokyo Institute of Technology |
| Akira Yokoi |
Lecturer, Obstetrics and Gynecology, Hospital, Nagoya University |
Description
While the medical technologies have been developed today、 there are still many diseases that have no initial symptoms and are difficult to detect and treat early. In this research、 we aim to acquire scientific basis for odor-mediated disease diagnosis by the odor of metabolites and establish a non-invasive odor sensing platform technology for early treatment of hard-to-find gynecological diseases by accelerating the development of comprehensive odor molecule analysis technology and artificial olfactory sensor technology that have been studied in the PRESTO research.
High Performance Data Science
Research Director
| Yusuke Matsui |
Lecturer, Graduate School of Information Science and Technology, The University of Tokyo |
Collaborators
| Daichi Amagata |
Assistant Professor, Graduate School of Information Science and Technology, Osaka University |
| Hiroaki Shiokawa |
Associate Professor, Center for Computational Sciences, University of Tsukuba |
| Mai Nishimura |
Senior Researcher, Research Administrative Division,OMRON SINIC X Corporation |
Description
Our research plan aims to develop a data science infrastructure for high-performance data science. Data science is an area that has attracted the most attention in the past decade, both in industry and academia. Accelerating the computation for data science is extremely important, both theoretically and practically. Fundamental operations in data science are attributed to basic mathematical operations such as search and clustering. Such mathematical operations are currently being developed independently in each field. In this research project, we will integrate acceleration techniques developed in various areas in the past. We then establish fast search techniques that we can apply to multiple modalities, including vector-quantized codes, arbitrary metric spaces, and graphs. Our methodologies will enable a quick search for massive data above the billion scale, regardless of the target modality.
An On-Device Learning Technology Enabling Both Adaptivity and Reliability
Research Director
| Hiroki Matsutani |
Professor, Faculty of Science and Engineering, Keio University |
Collaborators
| Tamao Okamoto |
Manager, Product Analysis Center, Panasonic Holdings Corporation |
| Kota Yoshida |
Assistant Professor, College of Science and Engineering, Ritsumeikan University |
Description
This work focuses on an on-device learning technology of neural networks on resource-limited IoT devices such as sensors and controllers and improves the adaptivity and reliability. Specifically, an autonomous retraining technique is enhanced by concept drift detection algorithms. Performance degradation due to unintended retraining and security attacks by malicious users are also addressed. The proposed techniques are demonstrated with real applications including electrical safety and equipment monitoring.
Computer-Aided Cytological Screening System by 3D AI and Data Flow Computer
Research Director
| Ken'ichi Morooka |
Professor, Faculty of Advanced Science and Technology, Kumamoto University |
Collaborators
| Eiji Ohno |
Visiting Researcher, Research Center for Life and Health Sciences, Kyoto Tachibana University |
| Hajime Nagahara |
Professor, Institute for Datability Science, Osaka University |
| Hideki Hashimoto |
General Manager, R&D Planning Division, Proassist, Ltd |
Description
The purpose of our project is to develop a cloud system for supporting a cytological screening by using 3D AI for identifying cells and data-flow computer.
