The paper of Blei et Lafferty published at ICML'06 implements a (quite complicated) variational inference (VI) technique for estimating the parameters of the Dynamic Topic Model, see:


I believe that an estimation driven by gibb's sampling can do the job. So my first question is: Do you think that GS can be competitive with VI for such complicated model? And the second question is: Does someone know a paper/software that already perform such inference? This can save me a lot of time.

Thank you very much,



Two recent papers are about it:

1) Arnab Bhadury, Jianfei Chen, Jun Zhu, and Shixia Liu. Scaling up Dynamic Topic Models, In Prof. of World Wide Web Conference (WWW), Montreal, Canada, 2016. (WWW 2016)

2) Scott W. Linderman*, Matthew J. Johnson*, Ryan P. Adams. Dependent multinomial models made easy: stick breaking with the Polya-Gamma augmentation. Neural Information Processing Systems (NIPS), 2015.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.