Parallel-tandem mode identification method and its uses in machine scent

Abstract

A discriminating method of parallel series patter and its application in the machinery smell, the characteristics is that regarding the big scale smell sorts and intension confirm problem as assortment problem, and then regarding as approximation problem, giving settling respectively for the mode which is made up of classifier layer and function approximation mode layer. It includes the four implementation steps: (1) n pieces one output neural net of classifier layer takes charge of ensuring the types of n kinds of smell, one to one correspondence. The training subset of a network is made up of all representative smell samples and the part apposition smell samples. (2) n pieces one output neural net of function approximation mode layer takes charge of ensuring the intensity of n kinds of smell, one to one correspondence. The training subset of a network is made up of all representative smell samples. (3) The suppositional balance of imbalance training subset. (4) n+1 pieces one output neural net puts up decision. By using the invention, the machinery olfactometer can solve the simultaneous estimation problem of thousands upon thousands smell types and intensity.

Description

technical field [0001] The invention is a pattern recognition method for large-scale learning problems and its application in real-time determination of large-scale odor categories and strengths by machine olfactory instruments, involving task decomposition, virtual balance of unbalanced sample sets, and parallel-serial neural network models The structure and parameter optimization of , the simultaneous determination method of odor category and intensity based on machine olfaction and its application. Background technique [0002] Spices and flavors, alcohol, cigarettes, tea, rice, wheat, cooking oil and other items have aroma as a quality inspection index. The current inspection "instrument" is the human nose, and the quality of aroma is described by "normal" , "pure", "fair", "dense" and other extremely vague terms, it is difficult to be scientific, objective and fair. The invention uses a plurality of gas sensors with overlapping performances to form an array to imitate ...

AI Technical Summary

Problems solved by technology

The classifier ensemble model and its combination strategy that appeared in the early 1990s opened up new ways to improve the generalization ability of classifiers. However, not only the rules such as maximum, minimum, average, product, sum, and majority voting cannot be directly applied to Combination of function approximation models, and the approximation ability of "weak" multivariate polynomial models such as linear and quadratic is not enough

Invention patent "A Method for Machine Olfactory Odor Recognition Based on Modular Combination Neural Network" (approval number: ZL03141537.7) transforms the concentration estimation problem into a classification problem, and does not involve the simultaneous estimation of odor category and concentration

Invention patent "Method for Simultaneous Determination of Odor Type and Intensity in Machine Sense of Smell" (Application No.: 200610023183.4, Publication No.: 1801136) regards the simultaneous determination of multiple odor categories and intensities as a multi-function approximation problem, and is approximated by multiple functions The model ensemble is solved one by one. The disadvantage is that the structure of multiple function approximation model ensembles is too complex, there are too many undetermined parameters, and the accuracy of classification and intensity estimation is not high enough.

Images