Gossip-Based Computation of a Gaussian Mixture Model for Distributed Multimedia Indexing

This paper deals with pattern recognition in a distributed computing context of the peer-to-peer type, that should be more and more interesting for multimedia data indexing and retrieval. Our goal is estimating of class-conditional probability densities, that take the form of Gaussian mixture models (GMM). Originally, we propagate GMMs in a decentralized fashion (gossip) in a network, and aggregate GMMs from various sources, through a technique that only involves little computation and that makes parsimonious usage of the network resource, as model parameters rather than data are transmitted. The aggregation is based on iterative optimization of an approximation of a KL divergence allowing closed-form computation between mixture models. Experimental results demonstrate the scheme to the case of speaker recognition.

[1]  Cordelia Schmid,et al.  Weakly Supervised Learning of Visual Models and Its Application to Content-Based Retrieval , 2004, International Journal of Computer Vision.

[2]  Patrick Gros,et al.  Robust content-based image searches for copyright protection , 2003, MMDB '03.

[3]  Nikos A. Vlassis,et al.  Newscast EM , 2004, NIPS.

[4]  Cordelia Schmid,et al.  Toward Category-Level Object Recognition , 2006, Toward Category-Level Object Recognition.

[5]  Douglas A. Reynolds,et al.  Speaker identification and verification using Gaussian mixture speaker models , 1995, Speech Commun..

[6]  H. Akaike.  A new look at the statistical model identification , 1974 .

[7]  David J. C. MacKay,et al.  Bayesian Interpolation , 1992, Neural Computation.

[8]  Anne-Marie Kermarrec,et al.  Epidemic information dissemination in distributed systems , 2004, Computer.

[9]  Johannes Gehrke,et al.  Gossip-based computation of aggregate information , 2003, 44th Annual IEEE Symposium on Foundations of Computer Science, 2003. Proceedings..

[10]  Sandhya Dwarkadas,et al.  On scaling latent semantic indexing for large peer-to-peer systems , 2004, SIGIR '04.

[11]  Andreas Henrich,et al.  Efficient content-based P2P image retrieval using peer content descriptions , 2003, IS&T/SPIE Electronic Imaging.

[12]  Lexing Xie,et al.  Slightly Supervised Learning of Part-Based Appearance Models , 2004, 2004 Conference on Computer Vision and Pattern Recognition Workshop.

[13]  Roger Mohr,et al.  Mixture densities for video objects recognition , 2000, Proceedings 15th International Conference on Pattern Recognition. ICPR-2000.

[14]  D. Milojicic,et al.  Peer-to-Peer Computing , 2010 .

[15]  Patrick Pérez,et al.  Nonparametric motion characterization using causal probabilistic models for video indexing and retrieval , 2002, IEEE Trans. Image Process..

[16]  Hagai Attias,et al.  A Variational Bayesian Framework for Graphical Models , 1999 .

[17]  Jacob Goldberger,et al.  Hierarchical Clustering of a Mixture Model , 2004, NIPS.

[18]  Christophe Biernacki,et al.  Choosing starting values for the EM algorithm for getting the highest likelihood in multivariate Gaussian mixture models , 2003, Comput. Stat. Data Anal..

[19]  Nikos A. Vlassis,et al.  Accelerated EM-based clustering of large data sets , 2006, Data Mining and Knowledge Discovery.

[20]  Nicu Sebe,et al.  Semisupervised learning of classifiers: theory, algorithms, and their application to human-computer interaction , 2004, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[21]  Robert D. Nowak,et al.  Distributed EM algorithms for density estimation and clustering in sensor networks , 2003, IEEE Trans. Signal Process..

[22]  Anne-Marie Kermarrec,et al.  From Epidemics to Distributed Computing , 2004 .

[23]  Stephen P. Boyd,et al.  Randomized gossip algorithms , 2006, IEEE Transactions on Information Theory.

[24]  Heekuck Oh,et al.  Neural Networks for Pattern Recognition , 1993, Adv. Comput..