Topic stability in LDA models

Asked Jun 01 '15 at 02:24

Active Jun 01 '15 at 02:26

Viewed 554 times

My major is bioinformatics and I want to use LDA (Latent Dirichlet Allocation) to explain histone code of a lot of genes.

I used the LDA model in my project, when I re-run my topic models with a different random seed, the topic terms seem to change dramatically.

I certainly do not want to my results to be an artifact of a random seed. Is there anything I can do to figure it out?

edited Jun 01 '15 at 02:26

TimoStaudinger

asked Jun 01 '15 at 02:24

Fiona_Wang

Can you give more details about the language and libraries you're using? – Daniel Compton Jun 01 '15 at 03:06
Have you made any progress on this? I have a similar problem. My solution has been to run the model many times and average the results. – Sam Weisenthal Jul 22 '15 at 22:42

0 Answers0