Artificial intelligence (AI) machine learning is used to perform full-spectrum identification on spectrum data produced by spectral measurement equipment (such as by Raman spectroscopy or near-infrared spectroscopy). Taking pesticide residue detection for example, the first step of the training process is to provide spectra corresponding to the crops that are pesticide-free or whose pesticide residue amounts are lower than the maximum residue level. The AI system learns from the spectra in a one-class support vector machine (one-class SVM) mode and generates a set of training models. The training models are then applied to spectral measurements of actual crop samples to calculate the outlying degree of the measurements and produce a value for quantifying the difference between an unknown sample and a sample that is known to meet the residue level requirement. The value produced reflects the risk of the amount of pesticide residues in a sample.

線上展網址：
https://tievirtual.twtm.com.tw/iframe/1561c0ea-d34f-4957-ba59-7a2883d540e7?group=3a1a21ad-59dd-4f93-9cbd-9b4a13294f3c&lang=en