Fast maximum margin matrix factorization for collaborative prediction Export

@inproceedings{citeulike:5160953, abstract = {Maximum Margin Matrix Factorization (MMMF) was recently suggested (Srebro et al., 2005) as a convex, infinite dimensional alternative to low-rank approximations and standard factor models. MMMF can be formulated as a semi-definite programming (SDP) and learned using standard SDP solvers. However, current SDP solvers can only handle MMMF problems on matrices of dimensionality up to a few hundred. Here, we investigate a direct gradient-based optimization method for MMMF and demonstrate it on large collaborative prediction problems. We compare against results obtained by Marlin (2004) and find that MMMF substantially outperforms all nine methods he tested.}, address = {New York, NY, USA}, author = {Rennie, Jasson D. M. and Srebro, Nathan}, booktitle = {ICML '05: Proceedings of the 22nd international conference on Machine learning}, citeulike-article-id = {5160953}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=1102351.1102441}, citeulike-linkout-1 = {http://dx.doi.org/10.1145/1102351.1102441}, doi = {10.1145/1102351.1102441}, isbn = {1-59593-180-5}, keywords = {factorization, hinge, loss, margin, matrix, maximum, smooth}, location = {Bonn, Germany}, pages = {713--719}, posted-at = {2009-11-03 18:35:05}, priority = {3}, publisher = {ACM}, title = {Fast maximum margin matrix factorization for collaborative prediction}, url = {http://dx.doi.org/10.1145/1102351.1102441}, year = {2005} }

TY - CONF ID - citeulike:5160953 L3 - citeulike-article-id:5160953 N2 - Maximum Margin Matrix Factorization (MMMF) was recently suggested (Srebro et al., 2005) as a convex, infinite dimensional alternative to low-rank approximations and standard factor models. MMMF can be formulated as a semi-definite programming (SDP) and learned using standard SDP solvers. However, current SDP solvers can only handle MMMF problems on matrices of dimensionality up to a few hundred. Here, we investigate a direct gradient-based optimization method for MMMF and demonstrate it on large collaborative prediction problems. We compare against results obtained by Marlin (2004) and find that MMMF substantially outperforms all nine methods he tested. CY - Bonn, Germany SN - 1-59593-180-5 AD - New York, NY, USA EP - 719 TI - Fast maximum margin matrix factorization for collaborative prediction PB - ACM T2 - ICML '05: Proceedings of the 22nd international conference on Machine learning SP - 713 KW - factorization KW - hinge KW - loss KW - margin KW - matrix KW - maximum KW - smooth AU - Rennie, Jasson AU - Srebro, Nathan PY - 2005/// UR - http://dx.doi.org/10.1145/1102351.1102441 ER -