TENSOR OBJECT CLASSIFICATION VIA MULTILINEAR DISCRIMINANT ANALYSIS NETWORK

Rui Zeng (1, 2) , Jiasong Wu (3, 1, 2) , Lotfi Senhadji (3, 1) , Huazhong Shu (1, 2)

1 CRIBS - Centre de Recherche en Information Biomédicale sino-français
2 LIST - Laboratory of Image Science and Technology [Nanjing]
3 LTSI - Laboratoire Traitement du Signal et de l'Image

This paper proposes an multilinear discriminant analysis network (MLDANet) for the recognition of multidimensional objects, knows as tensor objects. The MLDANet is a variation of linear discriminant analysis network (LDANet) and principal component analysis network (PCANet), both of which are the recently proposed deep learning algorithms. The MLDANet consists of three parts: 1) The encoder learned by MLDA from tensor data. 2) Features maps obtained from decoder. 3) The use of binary hashing and histogram for feature pooling. A learning algorithm for MLDANet is described. Evaluations on UCFll database indicate that the proposed MLDANet outperforms the PCANet, LDANet, MPCA+LDA, and MLDA in terms of classification for tensor objects.

Keywords: recognition; MLDANet; LDANet; Learning Algorithm; PCANet; Deep learning

Domains: Engineering Sciences [physics]; Signal and Image processing; Life Sciences [q-bio]; Bioengineering

Rui Zeng, Jiasong Wu, Lotfi Senhadji, Huazhong Shu. TENSOR OBJECT CLASSIFICATION VIA MULTILINEAR DISCRIMINANT ANALYSIS NETWORK. 40th IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Apr 2014, Brisbane, Australia. pp.1971--1975. ⟨hal-01380110⟩