Intelligent Sensory Evaluation Based on Support Vector Machines

Support vector machines (SVMs) are provided with great abilities of analyzing data with the characteristics of small sample-sets, high dimension, nonlinear, high noise. They are applicable to deal with machine learning problems of industries. The paper brought forward to taking advantage of multi-classification SVMs to evaluate the sensory qualities of products according to the data feature of such industries. The simulative experiments were done on the factual data-sample offered by a tobacco factory. The results validated the practical performance of SVMs learning models, which could satisfy the necessary of product designs.