This talk is based on the article: http://jmlr.org/papers/volume18/16-541/16-541.pdf
In a feature allocation model, each data point depends on a collection of unobserved latent features. For example, we might classify a corpus of texts by describing each document via a set of topics; the topics then determine a distribution over words for that document. In a Bayesian nonparametric setting, the Indian Buffet Process (IBP) is a popular prior in which the number of topics is unknown a priori. However, the IBP is static in that it does not account for the change in popularity of topics over time. This talk introduces the Wright-Fisher Indian Buffet Process (WF-IBP), a probabilistic model for collections of time-stamped documents. This is applied to develop a nonparametric focused topic model for collections of time-stamped text documents, and explore the full corpus of NIPS papers published from 1987 to 2015.
Bio:
Valerio Perrone is a final year PhD student at the Oxford-Warwick Statistics Programme (OxWaSP), working under the joint supervision of Professor Yee Whye Teh (Oxford), Dr. Dario Spanò (Warwick), and Dr. Paul Jenkins (Warwick). His research interests lie in the fields of Bayesian machine learning and deep learning. In particular, he is interested in developing algorithms for large-scale machine learning and Bayesian nonparametric models with realistic dependency structures. Applications of his work include topic modelling, recommender systems and population genetics.
#TuringSeminars
Add comment