| Project/Area Number |
17K14687
|
|---|
| Research Category |
Grant-in-Aid for Young Scientists (B)
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Communication/Network engineering
|
|---|
| Research Institution | Suzuka National College of Technology (2018-2019)
Anan National College of Technology (2017) |
|---|
Principal Investigator |
Ikuta Chihiro 鈴鹿工業高等専門学校, その他部局等, 講師 (70757319)
|
|---|
| Project Period (FY) |
2017-04-01 – 2020-03-31
|
| Project Status |
Completed (Fiscal Year 2019)
|
| Budget Amount *help |
¥2,080,000 (Direct Cost: ¥1,600,000、Indirect Cost: ¥480,000)
Fiscal Year 2019: ¥520,000 (Direct Cost: ¥400,000、Indirect Cost: ¥120,000)
Fiscal Year 2018: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
Fiscal Year 2017: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
|
|---|
| Keywords | ニューラルネットワーク / グリア細胞 / 深層学習 / パターン分類 / グリア / 人工グリアニューラルネットワーク / 畳み込みニューラルネットワーク / ドロップアウト / 機械学習 / ノイズ |
|---|
| Outline of Final Research Achievements |
In this study, novel artificial glial neural networks that include features of glial cell are developed. The glial cell had been considered to support cell of neuron. Recently, researchers discovered that the glial cell transmits signal by using various important ions. Based on these researches, the artificial glial neural network includes these new features of glial cell such as a calcium wave by glial cell and glial cell working to a synapse plasticity. In one of proposed model based on glial cell working to synaptic plasticity, the glial cell makes neuron clusters and decides a dropout ratio in each neuron cluster. By this work, the influence of proposed glial dropout is better than the influence of original dropout for training of the neural network. The proposed deep glial neural networks compared with existing deep neural network, and this study confirmed that the proposed model has a high ability of the training and generalization capability.
|
| Academic Significance and Societal Importance of the Research Achievements |
脳内に存在するグリア細胞に着目し人工ディープニューラルネットワークに応用する研究である．これまで，脳の高次情報処理はニューロンによるものと考えられてきたが本研究を通じてグリア細胞の働きの重要性がより認知され新たな研究の方向性を示すことができたと考えている．また，近年急速に導入が進んでいる人工知能の代表であるディープニューラルネットワークにグリア細胞の特徴を組み込むことによる性能向上を確認することができた．これは，今後のディープニューラルネットワークの発展にとって新たな方向性となると期待できる．
|
