2020 Fiscal Year Final Research Report
Elucidation of hub phenomenon occurring in large-scale data and its application to bio-medical data
| Project/Area Number |
16H02821
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Multimedia database
|
|---|
| Research Institution | Yamagata University (2017-2020)
National Institute of Genetics (2016) |
|---|
Principal Investigator |
Hara Kazuo 山形大学, 理学部, 准教授 (30467691)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
鈴木 郁美 長崎大学, 情報データ科学部, 准教授 (20637730)
|
|---|
| Project Period (FY) |
2016-04-01 – 2020-03-31
|
| Keywords | 近傍検索 |
|---|
| Outline of Final Research Achievements |
We have developed a method to suppress the appearance of hubs by converting the distance between data so that the data density is uniform. In particular, as an important graph construction method in graph-based semi-supervised learning, we proposed a method that does not have hubs and does not require an excessive reduction in the number of edges. Furthermore, when we investigated whether hubness occurred in bio-sequence data i.e., whether a specific sequence was similar to many other sequences, we confirmed that hubness occurred.
|
| Free Research Field |
情報学
|
| Academic Significance and Societal Importance of the Research Achievements |
Radovanovic et al. JMLR 2010によって「（グラフ上ではなく）空間上のハブ」に起因する問題が提起されて以来，ハブを解消または利用する方法（例えば新たなグラフィカルモデルやクラスタリング法）の開発，および，各ドメインタスクへの適用は，国際的な競争となりつつある．取り分け，音楽情報検索におけるハブを取り除く研究は，オーストリアの研究グループOFAIが世界をリードしている（Schnitzer et al. JMLR 2012）．本研究の成果は，医療生命系データにおけるハブの問題を，世界に先駆けて解決する土台となるものである．
|
