Title :
Multiple Instance Support Vector Machines with latent variable description
Author :
Jianjiang Lu ; Wei Li ; Jiabao Wang ; Yafei Zhang ; Yang Li ; Lei Bao
Author_Institution :
Coll. of Command Inf. Syst., PLA Univ. of Sci. & Technol., Nanjing, China
Abstract :
In this paper, the latent variable model is adopted to re-describe MI-SVM and its feature mapping variants. MI-SVM with latent variable description and the corresponding stochastic optimization learning algorithm are proposed. In the Musk and Corel datasets, the proposed algorithm achieves higher predicting accuracy and faster learning speed, with strong stability and robustness for parameters and noise.
Keywords :
learning (artificial intelligence); optimisation; stochastic processes; support vector machines; Corel datasets; MI-SVM; Musk datasets; feature mapping variants; latent variable description; multiple instance support vector machines; stochastic optimization learning algorithm; Image recognition; Irrigation; Noise; Radio frequency; Support vector machines; Latent variable models; Multiple instance learning; Stochastic gradient descent; Support vector machines;
Conference_Titel :
Fuzzy Systems and Knowledge Discovery (FSKD), 2013 10th International Conference on
Conference_Location :
Shenyang
DOI :
10.1109/FSKD.2013.6816236
