-
- Description:
SVDFeature is a toolkit developed by Apex Data & Knowledge Management Lab during the competition of KDDCup'11. It is designed to solve the feature-based matrix factorization efficiently. New models can be developed just by defining new features. The feature-based setting allows us to include many kinds of information into the model, making the model informative. Using the toolkit, we can easily incorporate information such as temporal dynamics, neighbourhood relationship, and hierarchical information into the model. Both learning to rank and rate prediction models are supported.
The toolkit is designed for facing large-scale dataset in CF problems for both rate prediction and ranking. During the training process, we don't have load all the training data into main memory. A double thread pipeline technique is used to speed up the fetching of training data to make the model training efficient and memory frugal.
- Changes to previous version:
No change in the toolkit, add a detailed non-trivial collaborative ranking experiment KDDCup'11 track2 using SVDFeature, getting state-of-art performance. The link to the experiment scripts are in our project page
- BibTeX Entry: Download
- Corresponding Paper BibTeX Entry: Download
- Supported Operating Systems: Linux, Windows, Unix
- Data Formats: Ascii
- Tags: Large Scale Learning, Collaborative Filtering, Contextual Aware Recommendation, Kddcup2011
- Archive: download here
Comments
No one has posted any comments yet. Perhaps you'd like to be the first?
Leave a comment
You must be logged in to post comments.