Mahalanobis distance based similarity regression learning of NIRS for quality assurance of tobacco product with different variable selection methods

An improved sample container cup and its effect on NIRS of tobacco smoke for quality stability estimation

“…The product stability estimation method presented in this paper is generally applicable to most tobacco brands with a single tobacco leaves source. However, for cigarettes made from a mixture of multiple types of tobacco leaves, it is recommended to combine the product stability estimation method described in this paper with other machine learning related algorithms to enhance accuracy, as detailed in our related paper [1] .…”

“…Traditionally, tobacco smoke estimation has relied on human investigators' smoking experience in tobacco companies, which can be easily influenced by subjective factors. In our previous paper, a method to measure the cigarettes’ periodical product stability was introduced [1] . While numerous methods for tobacco quality control and chemical stability exist, many of them focus on tobacco leaves due to their ease of collection and measurement compared to tobacco smoke [2] , [3] , [4] , [5] .…”

Two-level fusion of multi-sensor information for compressor blade crack detection based on self-attention mechanism

“…In addition, for covariance R , the row number must be more than the column number, which means that the group must have abundant samples. 31 Generally, the more sample numbers, the better similarity analysis.…”

Feasibility of an NIR spectral calibration transfer algorithm based on optimized feature variables to predict tobacco samples in different states

“…NIR spectroscopy is extensively used in the detection and quality control of complex natural products, such as agricultural products, 7,8 Chinese herbal medicines 9,10 and petroleum, 11,12 due to its recognized advantages of being rapid, efficient, non-destructive and green. 13 Moreover, in the tobacco industry, [14][15][16] it has been used successfully for a long time. Affluent structural information is obtained from the NIR spectrum; then, multiple regression techniques, such as partial least-squares regression (PLSR), are applied to establish the link between the spectra and the target property reference values.…”