We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
ディープラーニング等高線HPLC 法を用いた食用きのこ識別に関する研究.
- Authors
北尾修平; 森山祐羽; 高山卓大; 井之上浩一
- Abstract
For the identification of various mushrooms regarding to food, it has been used to evaluate these morphologic and/or genetic decisions by these researchers and experts. However, these judgments require expertness and know-how on uncertainty approach as human error. In this study, we developed the deep learning contour HPLC method for the high-probability decision of various mushroom samples based on convolutional neural network (CNN). HPLC condition is TSKgel ODS-100V with 0.1% formic acid in water/acetonitrile at 1.0 mL/min and detectable 190-500 nm for contour chromatograms. The sample preparation was used of extraction with methanol. Contour chromatograms of extracted mushroom based on this HPLC measurement were transformed to NumPy form and replicated by 2D-image for CNN by Python. This result showed that accuracy is 94.9% (n=117, 13 kinds). Thus, it is useful that this novel approach with contour HPLC and CNN indicates the establishment of various natural components based on deep learning.
- Subjects
CONVOLUTIONAL neural networks; DEEP learning; HUMAN error; HIGH performance liquid chromatography; RESEARCH personnel; FORMIC acid
- Publication
Japanese Journal of Food Chemistry & Safety, 2023, Vol 30, Issue 3, p128
- ISSN
1341-2094
- Publication type
Article