Scalable Bayesian Modelling of Paired Symbols
This work addresses the challenge of scalable probabilistic modeling for large datasets in domains like recommendation systems, though it appears incremental as it builds on existing variational methods.
The authors tackled the problem of modeling pairs of symbols from large vocabularies by proposing a scalable Bayesian approach that uses variational bounding and site-independent bounds for inference, achieving state-of-the-art results on real-world movie viewing data.
We present a novel, scalable and Bayesian approach to modelling the occurrence of pairs of symbols (i,j) drawn from a large vocabulary. Observed pairs are assumed to be generated by a simple popularity based selection process followed by censoring using a preference function. By basing inference on the well-founded principle of variational bounding, and using new site-independent bounds, we show how a scalable inference procedure can be obtained for large data sets. State of the art results are presented on real-world movie viewing data.