| Project/Area Number |
16K16262
|
|---|
| Research Category |
Grant-in-Aid for Young Scientists (B)
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Eating habits
|
|---|
| Research Institution | Tokyo University of Marine Science and Technology |
|---|
Principal Investigator |
Shibata Mario 東京海洋大学, 学術研究院, 助教 (40590360)
|
|---|
| Project Period (FY) |
2016-04-01 – 2018-03-31
|
| Project Status |
Completed (Fiscal Year 2017)
|
| Budget Amount *help |
¥4,030,000 (Direct Cost: ¥3,100,000、Indirect Cost: ¥930,000)
Fiscal Year 2017: ¥650,000 (Direct Cost: ¥500,000、Indirect Cost: ¥150,000)
Fiscal Year 2016: ¥3,380,000 (Direct Cost: ¥2,600,000、Indirect Cost: ¥780,000)
|
|---|
| Keywords | 蛍光指紋 / 非破壊計測 / 多変量解析 / 魚肉 / 魚 / ニューラルネットワーク / 鮮度 / 機械学習 |
|---|
| Outline of Final Research Achievements |
Predictive models for of freshness or umami from the fluorescence fingerprint, nucleic acid-related substance and amino acid contents of tuna meat (Thunnus obesus) were constructed by changing the thawing and freezing conditions of the samples. Fluorescence fingerprint and K-value were set as an explanatory variables and as a response variable. As a result, PLS regression analysis showed that the coefficient of determination for the validation data was 0.90 with 8 latent variables. On the other hand, by optimizing parameters such as the number of nodes in the mid layer and the performance ratio, the neural network model showed the best performance with coefficient of determination of 0.97. It is suggested that neural network may be effective for construction of predictive models from fluorescent fingerprints.
|
