Collaborative Filtering with Label Consistent Restricted Boltzmann Machine
This work addresses the need for more accurate recommendations by enhancing an unsupervised RBM approach with supervision, though it is incremental as it builds on a known but underutilized method.
The paper tackled the problem of improving collaborative filtering in recommender systems by incorporating user demographic information and item metadata into a Restricted Boltzmann Machine (RBM) framework, resulting in significant improvements over existing RBM-based methods and achieving performance comparable to state-of-the-art latent factor models.
The possibility of employing restricted Boltzmann machine (RBM) for collaborative filtering has been known for about a decade. However, there has been hardly any work on this topic since 2007. This work revisits the application of RBM in recommender systems. RBM based collaborative filtering only used the rating information; this is an unsupervised architecture. This work adds supervision by exploiting user demographic information and item metadata. A network is learned from the representation layer to the labels (metadata). The proposed label consistent RBM formulation improves significantly on the existing RBM based approach and yield results at par with the state-of-the-art latent factor based models.