Wow, it's almost a year since last update. What have I done in the last year ?
*SVD++. This is similar to the SVD++ described by BellKor, or the transdusive MF described by Gravity.
*Expanded KNN, combinations of movie/viewer and raw/residue/binary, and much faster too (raw movie correlation calcuation takes less than 2 minutes now). For all my KNNs, I no longer save huge correlation files, instead the program just calculates correlations and predictions in one pass and saves predictions only.
*Recently got basic RBM working. This is the most confusing model for me. Basically I got it working thru trial and error. I still don't really understand it, and I have yet to make it work on residues of other predictors.
*Read a lot of research papers. I actually spent more time reading papers than writing code.
6 comments:
I was wondering if you're planning on releasing parts of your code, or something of the sort.
I'm particularly interested in the implementation of KNN.
Welcome back! Your blog has been a big help to me, I really appreciate it.
Actually, I'd like to release the source code, but it takes effort.
To make it useable and hopefully understandable, I need to release quite a few support programs that build my whole framework, and I haven't touched those little programs for two years. So I have to spend time looking for them, reviewing and repeating the framework building steps, and write some documentation so people can use them. Well, I just don't have the motivation to spend that much effort on it.
I'm curious as to how much the RMB has improved your score?
Improvement from RBM is actually pretty disappointing: about 0.0015.
I'm interested in how you got the RBM to work. The original RBM-Netflix paper contains errors and is not specific enough to duplicate its results.
Post a Comment