Infinite Hidden Relational Models

@inproceedings{citeulike:1816299, abstract = {Relational learning analyzes the probabilistic constraints between the attributes of entities and relationships. We extend the expressiveness of relational models by introducing for each entity (or object) an infinitedimensional latent variable as part of a Dirichlet process (DP) mixture model. We discuss inference in the model, which is based on a DP Gibbs sampler, i.e., the Chinese restaurant process. We extended the Chinese restaurant process to be applicable to relational modeling. We discuss how information is propagated in the network of latent variables, reducing the necessity for extensive structural learning. In the context of a recommendation engine our approach realizes a principled solution for recommendations based on features of items, features of users and relational information. Our approach is evaluated in three applications: a recommendation system based on the Movie- Lens data set, the prediction of gene function using relational information and a medical recommendation system.}, author = {Xu, Zhao and Tresp, Volker and Yu, Kai and Kriegel, Hans-Peter }, booktitle = {Proc. 22nd Conf. on Uncertainty in Artificial Intelligence}, citeulike-article-id = {1816299}, keywords = {machine-learning}, posted-at = {2008-05-13 16:12:38}, priority = {2}, title = {Infinite Hidden Relational Models}, url = {http://www.dbs.ifi.lmu.de/Publikationen/Papers/ihrm\_uai06\_227.pdf}, year = {2006} }

TY - CONF ID - citeulike:1816299 L3 - citeulike-article-id:1816299 TI - Infinite Hidden Relational Models T2 - Proc. 22nd Conf. on Uncertainty in Artificial Intelligence N2 - Relational learning analyzes the probabilistic constraints between the attributes of entities and relationships. We extend the expressiveness of relational models by introducing for each entity (or object) an infinitedimensional latent variable as part of a Dirichlet process (DP) mixture model. We discuss inference in the model, which is based on a DP Gibbs sampler, i.e., the Chinese restaurant process. We extended the Chinese restaurant process to be applicable to relational modeling. We discuss how information is propagated in the network of latent variables, reducing the necessity for extensive structural learning. In the context of a recommendation engine our approach realizes a principled solution for recommendations based on features of items, features of users and relational information. Our approach is evaluated in three applications: a recommendation system based on the Movie- Lens data set, the prediction of gene function using relational information and a medical recommendation system. KW - machine-learning AU - Xu, Zhao AU - Tresp, Volker AU - Yu, Kai AU - Kriegel, Hans-Peter PY - 2006/// UR - http://www.dbs.ifi.lmu.de/Publikationen/Papers/ihrm_uai06_227.pdf ER -