Bayesian Optimization for Collaborative Filtering with MLlib

Re:Bayesian Optimization for Collaborative Filtering with MLlib by a1k0n on Tuesday August 09, 2016 @05:56PM (#12257999)

So SigOpt was tuning rank (number of latent factors), number of iterations to run the algorithm (in my experience alternating least squares generally converges within 10-20 iterations, but there'd be no downside to running it longer unless it's overfitting), and the regularization strength.

What optimal parameters did it find for these?

[ Parent | on HN ]

Re:Bayesian Optimization for Collaborative Filtering with MLlib by Zephyr314 on Tuesday August 09, 2016 @06:15PM (#12258099)

Great question. As you point out, all of these parameters (rank, num iterations and the actual reg. term) can in some way contribute to the regularization of the reconstruction of the ratings matrix. We thought it would be interesting to include all of them and here are the optimal parameters SigOpt found for this experiment:

rank = 36

numIter = 30

log_lambda = -2.90405347693

[ Parent | on HN ]