可追溯性
人工智能
多元统计
计算机科学
算法
模式识别(心理学)
数据挖掘
机器学习
软件工程
作者
Peng Chen,Chenghao Fei,Rao Fu,Xiaoyan Xiao,Yuwen Qin,Xiaoman Li,Zhijun Guo,Jianmin Huang,De Ji,Lang Lin,Tulin Lu,Qiaosheng Guo,Lianlin Su
标识
DOI:10.1016/j.foodchem.2024.140350
摘要
This study collected multidimensional feature data such as spectra, texture, and component contents of Polygonati Rhizoma from different origins and varieties (Polygonatum kingianum Coll. et Hemsl from Yunnan and Guizhou; Polygonatum cyrtonema Hua from Anhui and Jiangxi; Polygonatum sibiricum Red from Hunan). Multivariate statistical analysis was used to select 39 characteristic factors for distinguishing PR origins and 14 characteristic factors for discriminating PR varieties (VIP > 1 and P < 0.05). In addition, by combining multivariate statistical analysis with a deep belief network (DBN) classification algorithm, a novel artificial intelligence algorithm was developed and optimized. Compared to traditional discriminant analysis methods, the accuracy of this new approach was significantly improved, achieving a 100% discrimination rate for PR varieties and a 100% accuracy rate for tracing the origin of PR. This research provides a reference and data support for constructing intelligent algorithms based on multidimensional data fusion, to achieve food variety discrimination and origin tracing.
科研通智能强力驱动
Strongly Powered by AbleSci AI