Abstract
We present a fast variational Bayesian algorithm for performing non-negative matrix factorisation and tri-factorisation. We show that our approach achieves faster convergence per iteration and timestep (wall-clock) than Gibbs sampling and non-probabilistic approaches, and do not require additional samples to estimate the posterior. We show that in particular for matrix tri-factorisation convergence is difficult, but our variational Bayesian approach offers a fast solution, allowing the tri-factorisation approach to be used more effectively.
Original language | English |
---|---|
Publication date | 9 Dec 2016 |
Publication status | Published - 9 Dec 2016 |
Event | NIPS 2016: Advances in Approximate Bayesian Inference Workshop - Room 112, Centre Convencions Internacional Barcelona, Barcelona, Spain Duration: 9 Dec 2016 → 9 Dec 2016 http://approximateinference.org |
Workshop
Workshop | NIPS 2016 |
---|---|
Location | Room 112, Centre Convencions Internacional Barcelona |
Country/Territory | Spain |
City | Barcelona |
Period | 09/12/2016 → 09/12/2016 |
Internet address |
Keywords
- variational Bayesian algorithm
- non-negative matrix factorization
- matrix tri-factorisation
- fast convergence
- posterior estimation