无限最大间隔Beta过程因子分析模型

doi:10.3969/j.issn.1001-2400.2016.03.003

Abstract

Abstract:

An infinite max-margin Beta process factor analysis (IMMBPFA) model is developed to deal with the classification problem on multimodal data. In this model, BPFA is utilized to capture the latent feature of data. With the idea of mixture experts, IMMBPFA divides the data into ‘infinite’ clusters via the Dirichlet process (DP) mixture model in the low-dimensional latent space and meanwhile learns a linear max-margin classifier on each cluster to construct a complex nonlinear classifier. Since the proposed model jointly learns BPFA, clustering and max-margin classifier in a unified Bayesian framework, it exhibits superior performance in both data description and discrimination. With the help of nonparametric Bayesian inference and the Gibbs sampler, we avoid the model selection problem and can estimate the parameters simply and effectively. Based on the experimental data obtained from Benchmark and measured radar high resolution range profile (HRRP) dataset, the effectiveness of proposed method is validated.

Key words: Beta process factor analysis (BPFA), mixture-of-experts, Dirichlet process mixture(DPM) model, max-margin classifiers

ZHANG Xuefeng;CHEN Bo;WANG Penghui;WEN Wei;LIU Hongwei. Infinite max-margin Beta process factor analysis model[J].Journal of Xidian University, 2016, 43(3): 13-18+72.

References

［1］ TUNC B, DAGLI V, GOKMEN M. Class Dependent Factor Analysis and Its Application to Face Recognition ［J］. Pattern Recognition, 2012, 45(12): 4092-4102.
［2］ DU L, LIU H W, BO Z. Radar HRRP Statistical Recognition: Parametric Model and Model Selection ［J］. IEEE Transactions on Signal Processing, 2008, 56 (5): 1931-1943.
［3］ DU L, LIU H W, WANG P H, et al. Noise Robust Radar HRRP Target Recognition Based on Multitask Factor Analysis with Small Training Data Size ［J］. IEEE Transactions on Signal Processing, 2012, 60(7): 3546-3559.
［4］ BISHOP C. Pattern Recognition and Machine Learning ［M］. New York: Springer Science+Business Media, 2007.
［5］ POLSON N G, SCOTT S L. Data Augmentation for Support Vector Machines ［J］. Bayesian Analysis, 2011, 6(1): 1-24.
［6］ ZHU J, AHMED A, XING E P. MedLDA: Maximum Margin Supervised Topic Model ［J］. The Journal of Machine Learning Research, 2012, 13(1): 2237-2278.
［7］ ZHU J, CHEN N, PERKINS H, et al. Gibbs Max-margin Topic Models with Data Augmentation ［J］. The Journal of Machine Learning Research, 2014, 15(1): 1073-1110.
［8］ BLEI D M, JORDAN M I. Variational Inference for Dirichlet Process Mixtures ［J］. Bayesian Analysis, 2006, 1(1):121-144.
［9］ ZHU J, CHEN N, XING E P. Bayesian Inference with Posterior Regularization and Applications to Infinite Latent SVMs ［J］. The Journal of Machine Learning Research, 2014, 15(1): 1799-1847.
［10］ ZHU J, CHEN N, XING E P. Infinite SVM: a Dirichlet Process Mixture of Large-margin Kernel Machines ［C］//Proceedings of the 28th International Conference on Machine Learning. New York: ACM, 2011: 617-624.
［11］ SHAHBABA B, NEAL R. Nonlinear Models Using Dirichlet Process Mixtures ［J］. The Journal of Machine Learning Research, 2009, 10(4): 1829-1850.
［12］张学峰, 陈渤, 王鹏辉, 等. 一种基于Dirichlet过程隐变量支撑向量机模型的目标识别方法［J］. 电子与信息学报, 2015, 37(1): 29-36.
ZHANG Xuefeng, CHEN Bo, WANG Penghui, et al. A Target Recognition Method Based on Dirichlet Process Latent Variable Support Vector Machine Model ［J］. Journal of Electronics ＆ Information Technology, 2015, 37(1): 29-36
［13］ PAISLEY J, CARIN L. Nonparametric Factor Analysis with Beta Process Priors ［C］//Proceedings of the 26th International Conference on Machine Learning. New York: ACM, 2009: 777-784.
［14］ FERGUSON T S. A Bayesian Analysis of Some Nonparametric Problems ［J］. The Annals of Statistics, 1973, 1(2): 209-230.
［15］ SETHURAMAN J. A Constructive Definition of Dirichlet Priors ［J］. Statistica Sinica, 1994, 4(2): 639-650.

Infinite max-margin Beta process factor analysis model

PDF (PC)

Like

Knowledge

Abstract

Cite this article

share this article

References

Related Articles 0

Metrics

Comments

Recommended 10