Prof. Ljupco Kocarev
Research Centre for Energy, Informatics and Materials
Macedonian Academy of Science and Arts
Nonparametric Bayesian methods for networks
For most of the Bayesian methods data is represented by exchangeable sequences of observations. An infinite exchangeable sequence is strictly stationary and, therefore, a law of large numbers in the form of Birkhoff–Khinchin theorem applies. The close relationship between exchangeable sequences of random variables and the i.i.d. will be described as well as its connection to Bruno de Finetti’s development of predictive inference and to Bayesian statistics. Example models will be reviewed; applications of such models include collaborative filtering, link prediction, and graph and network analysis.