2022 Fiscal Year Final Research Report
Classification methods taking account clarity of characteristics in biological sounds.
| Project/Area Number |
20K12045
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 62010:Life, health and medical informatics-related
|
|---|
| Research Institution | Nagasaki University |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
高田 寛之 長崎大学, 情報データ科学部, 助教 (10297616)
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Keywords | 肺音 / 泣き声 / 識別 |
|---|
| Outline of Final Research Achievements |
In this study, we aimed to provide subjects with intuitive and easy-to-understand recognition results (detection classes) with higher accuracy based on the clarity of characteristic sounds in the biomedical sounds. We conducted this research through two topics, "detection of adventitious sounds in auscultation sounds (lung sounds)" and "discrimination of emotions contained in infant crying" to obtain general results.The classification classes were designed by considering the subjective easiness of detecting the characteristic sounds as clarity. We found that it is intuitively easy to understand for the subjects by classifying into a class with high clarity and other class. In addition, we developed the increasing method of training data for high accuracy using machine learning approach, effective machine learning methods and its acoustic input parameters.
|
| Free Research Field |
生体音処理
|
| Academic Significance and Societal Importance of the Research Achievements |
生体音に含まれる特徴的な情報の識別では，人間の主観に応じた判断結果を必要とする場合が多い．しかしながら，識別対象とする識別クラスの設定に関しては，従来は多くの場合，識別されるクラスが Top-down に，かつ，決定論的に定められるものであり，利用者に必ずしも有益な情報を与えていない場合も存在した．この問題を解決するために，本研究では人間の主観に応じた新たな識別クラスを提示することを検討したことに意義がある．また，識別クラスの増加に伴い機械学習が学習データの不足により，学習不能となることを防ぐことができる機械学習法と学習データの増強法の一例を示せたことに意義がある．
|
