Many kinds of data can be represented as a network or graph. It is crucial toinfer the latent structure underlying such a network and to predict unobservedlinks in the network. Mixed Membership Stochastic Blockmodel (MMSB) is apromising model for network data. Latent variables and unknown parameters inMMSB have been estimated through Bayesian inference with the entire network;however, it is important to estimate them online for evolving networks. In thispaper, we first develop online inference methods for MMSB through sequentialMonte Carlo methods, also known as particle filters. We then extend them fortime-evolving networks, taking into account the temporal dependency of thenetwork structure. We demonstrate through experiments that the time-dependentparticle filter outperformed several baselines in terms of predictionperformance in an online condition.
展开▼
