本文へジャンプ

成果報告書詳細
管理番号20190000000260
タイトル*平成30年度中間年報 NEDOプロジェクトを核とした人材育成、産学連携等の総合的展開/実データで学ぶ人工知能講座
公開日2019/6/25
報告書年度2018 - 2018
委託先名国立大学法人大阪大学
プロジェクト番号P06046
部署名ロボット・AI部
和文要約
英文要約Title: Educational program to learn Artificial Intelligence with actual data/Comprehensive development of human resources development and industry-academia collaboration centered on NEDO project (FY2017-2019) FY2018 Annual Report

Course in Osaka University
This course provides lectures of fundamentals of Artificial Intelligence, and learns from real-world data of brain image data and cell data. This course had 32 participants in the spring-and-summer semester, and 30 participants in the fall-and-winter semester of FY2018.

Lectures in the spring-and-summer semester
- Intelligence and Learning: Artificial intelligence and machine learning
- Big Data Engineering: Data mining and machine learning for big data analysis
- Brain Function Measurement: Neuro-science and data analysis

Lectures in the fall-and-winter semester
- Image Recognition: Pattern recognition for images
- Robot Vision: Computer vision technology for machine vision functions
- Knowledge Informatics: Machine learning and data mining for data analysis

Placement test was held before the lectures, and we offered the participants who did not satisfy the condition to take lectures of The Open University of Japan for extra class. The above three lectures provide fundamental machine learning and pattern recognition, afterward real-common-data exercise related to the lectures was conducted.

Real-common-data exercise
This year we developed a new machine learning prototyping framework which lets students build a learning machine from scratch. By using model, data, and algorithm objects as atomic elements, students translate abstract concepts into precise formulations (formula + coding). In the second half of the exercise class, students listened to classes and did the following tasks, classification of MNIST data by MLP and CNN, classification of images of cancer and normal cells with a famous model of ILSVRC, segmentation using U-Net, etc.

Exchange meeting between Osaka University and Tokyo University
This meeting was heled on July 28th, 2018 (spring-and-summer semester) and March 22nd, 2019 (fall-and-winter semester) to promote interaction of people of each participants, professors from both Universities and research institutions.

Related research
Collecting real-common-data for education:
Main purpose of this research is building large scale multimodal-data-set that includes bio signal from brain function, sleep, feelings, health and mental condition. Subjects continue measuring bio signal, recording environmental sound and answering some questionnaire for one month. In addition, daily feeling and mental condition by music stimulation can be measured by using MRI and MEG.
This year, we have succeeded to construct protocols and an environment to conduct the experiments continuously. We have collected totally 125 subjects (89 in FY2018) of sleep data, 35 subjects (19 in FY2018) of MRI and MEG data.
ダウンロード成果報告書データベース(ユーザ登録必須)から、ダウンロードしてください。

▲トップに戻る