Fast-Learning Adaptive-Subspace Self-Organizing Map: An Application to Saliency-Based Invariant Image Feature Construction

Huicheng Zheng (1) , Grégoire Lefebvre (2) , Christophe Laurent (2)

1 SYSU - Sun Yat-sen University [Guangzhou]
2 France Telecom Division R&D

The Adaptive-Subspace Self-Organizing Map (AS-SOM) is useful for invariant feature generation and visualization. However, the learning procedure of the ASSOM is slow. In this paper, two fast implementations of the ASSOM are proposed to boost ASSOM learning based on insightful discussions of the basis rotation operator of ASSOM. We reveal the objective function approximately maximized by the classical rotation operator. We then explore a sequence of two schemes to apply the proposed ASSOM implementations to saliency-based invariant feature construction for image classification. In the first scheme, a cumulative activity map computed from a single ASSOM is used as descriptor of the input image. In the second scheme, we use one ASSOM for each image category and a joint cumulative activity map is calculated as the descriptor. Both schemes are evaluated on a subset of the Corel photo base with 10 classes. The multi-ASSOM scheme is favored. It is also applied to adult image filtering and shows promising results.

Mots clés: Adaptive-Subspace Self-Organizing Map; feature construction; adult image filtering; image classification

Domaines: Vision par ordinateur et reconnaissance de formes [cs.CV]; Intelligence artificielle [cs.AI]; Apprentissage [cs.LG]; Traitement du signal et de l'image [eess.SP]

Huicheng Zheng, Grégoire Lefebvre, Christophe Laurent. Fast-Learning Adaptive-Subspace Self-Organizing Map: An Application to Saliency-Based Invariant Image Feature Construction. IEEE Transactions on Neural Networks, 2008, ⟨10.1109/TNN.2007.911741⟩. ⟨hal-01216103⟩