MULTITARGET CLASSIFICATION PATTERN-RECOGNITION APPLIED TO COMPUTER-AIDED MATERIALS DESIGN

A multi-target classification pattern recognition method based on the principal component analysis (PCA) has been proposed for computer-aide d materials design considering the multiple specialties of materials. The designed sample with probable optimal specialties is determined in such way that its representing point should be at the optimal region in the PC sub-space, where most optimal samples can be discriminated f rom nonoptimal samples by means of features. The experimental paramete rs of the sample represented by this point can be obtained using a non -linear inverse mapping method from the PC sub-space to the original s pace. Based on the information provided using the method, several samp les of V-PTC materials with optimal multiple specialties are synthesiz ed.