Download now Free registration required
A novel approach proposed to solve the supervised dimensionality reduction problem by using tensor objects. MPCA used for unsupervised dimension reduction. Then Discriminant Analysis with TEnsor Representation- (DATER) is proposed to find the best subspaces. It's important that both of those algorithms work in tensor space so the structure of the objects never broke. These algorithms are avoiding the curse of dimensionality by using higher-order tensors. At the end, the comprehensive experiments are provided on CMU-PIE, FERET databases.
- Format: PDF
- Size: 285.34 KB