Local Discriminant Embedding with Tensor Representation

@inproceedings{citeulike:2518626, abstract = {We present a subspace learning method, called local discriminant embedding with tensor representation (LDET), that addresses simultaneously the generalization and data representation problems in subspace learning. LDET learns multiple interrelated subspaces for obtaining a lower-dimensional embedding by incorporating both class label information and neighborhood information. By encoding each object as a second- or higher-order tensor, LDET can capture higher-order structures in the data without requiring a large sample size. Extensive empirical studies have been performed to compare LDET with a second- or third-order tensor representation and the original LDE on their face recognition performance. Not only does LDET have a lower computational complexity than LDE, but LDET is also superior to LDE in terms of its recognition accuracy}, author = {Xia, Jian and Yeung, Dit-Yan and Dai, Guang }, booktitle = {Image Processing, 2006 IEEE International Conference on}, citeulike-article-id = {2518626}, doi = {http://dx.doi.org/10.1109/ICIP.2006.312627}, journal = {Image Processing, 2006 IEEE International Conference on}, keywords = {metric-learning, tensor}, pages = {929--932}, posted-at = {2008-03-12 08:26:22}, priority = {2}, title = {Local Discriminant Embedding with Tensor Representation}, url = {http://dx.doi.org/10.1109/ICIP.2006.312627}, year = {2006} }

TY - CONF ID - citeulike:2518626 L3 - citeulike-article-id:2518626 N2 - We present a subspace learning method, called local discriminant embedding with tensor representation (LDET), that addresses simultaneously the generalization and data representation problems in subspace learning. LDET learns multiple interrelated subspaces for obtaining a lower-dimensional embedding by incorporating both class label information and neighborhood information. By encoding each object as a second- or higher-order tensor, LDET can capture higher-order structures in the data without requiring a large sample size. Extensive empirical studies have been performed to compare LDET with a second- or third-order tensor representation and the original LDE on their face recognition performance. Not only does LDET have a lower computational complexity than LDE, but LDET is also superior to LDE in terms of its recognition accuracy JF - Image Processing, 2006 IEEE International Conference on EP - 932 TI - Local Discriminant Embedding with Tensor Representation T2 - Image Processing, 2006 IEEE International Conference on SP - 929 KW - metric-learning KW - tensor AU - Xia, Jian AU - Yeung, Dit-Yan AU - Dai, Guang PY - 2006/// UR - http://dx.doi.org/10.1109/ICIP.2006.312627 ER -