Development and application of algorithms for information collection from learning targets
| Project/Area Number |
16K16121
|
|---|
| Research Category |
Grant-in-Aid for Young Scientists (B)
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | Nihon University (2018)
Institute of Physical and Chemical Research (2016-2017) |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2016-04-01 – 2019-03-31
|
| Project Status |
Completed (Fiscal Year 2018)
|
| Budget Amount *help |
¥3,900,000 (Direct Cost: ¥3,000,000、Indirect Cost: ¥900,000)
Fiscal Year 2018: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2017: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2016: ¥1,820,000 (Direct Cost: ¥1,400,000、Indirect Cost: ¥420,000)
|
|---|
| Keywords | 情報量最大化 / ニューラルネットワーク / 統計力学 / 強化学習 / リカレント神経回路 / ゲーム理論 / 非線形力学 / 非線形力学系 / 機械学習 |
|---|
| Outline of Final Research Achievements |
In this project, the researcher developed a new algorithm that makes a learner actively collect information about the target of learning and that is applicable even if the target itself learns and changes its strategy. The performance of the new algorithm was tested by theoretically analyzing the behaviour of the algorithm or by implementing the algorithm with neural networks in computer simulations. From these analyses, novel utility of information collection was discovered. In the simulations, the performance of the algorithm turned out to largely depend on the ability of the neural network to remember the time series of events in learning. Then, the researcher developed a new theory about the behaviour of neural networks and found that the theory allows us to design neural networks that efficiently remember time series. These results laid theoretical and technical foundations for information collection from learning targets.
|
| Academic Significance and Societal Importance of the Research Achievements |
我々人間が日常的に経験しているように、どのような種類の学習であっても、学習対象について情報収集することが効率的な学習の鍵となる。学習対象自体が知能を持たず変化しない場合については、コンピューターが自動的に情報収集するためのアルゴリズムが研究されてきたが、学習対象自体が知能を持ち学習し変化する場合についてはあまり研究されてこなかった。研究代表者は後者の場合にも適用できるコンピューターアルゴリズムを導き、その有用性を示し、応用面での技術的なブレイクスルーを得た。この成果を基盤として、人工知能の様々な学習機能が加速され、絶えず変化する環境に適応できるものになっていくことが期待される。
|
Report
(4 results)
Research Products
(6 results)
