mloss.org BMRMhttp://mloss.orgUpdates and additions to BMRMenFri, 08 May 2009 08:08:20 -0000BMRM 2.1http://mloss.org/software/view/187/<html><p>BMRM is an open source, modular and scalable convex solver for many machine learning problems cast in the form of regularized risk minimization problem. It is "modular" because the (problem-specific) loss function module is decoupled from the (regularization-specific) optimization module (e.g. quadratic programming or linear programming solvers), thus shorten the time to implement/prototype solutions to new problems. Besides, the decoupling leads to easier parallelization of the loss function computation. </p> <p>At the moment, BMRM can solve the following problems: </p> <ul> <li> Binary classification </li> <li> Hinge </li> <li> Squared hinge </li> <li> Huber-hinge </li> <li> Logistic regression </li> <li> Exponential </li> <li> ROC Score </li> <li> Fbeta Score </li> <li> Univariate regression </li> <li> Epsilon-insensitive </li> <li> Huber robust </li> <li> Least Mean Squares </li> <li> Least Absolute Deviation </li> <li> Novelty detection (1-class SVM) </li> <li> Quantile regression </li> <li> Poisson regression </li> <li> Ranking </li> <li> NDCG (normalized discounted cummulative gain) </li> <li> Graph Matching </li> <li> Sequence Segmentation and Classification </li> </ul> <p>along with either L1 or L2 regularizer. Also, users can add new loss function for problems with structured input and output variables. </p></html>choon hui teo, alex smola, s v n vishwanathan, quoc leFri, 08 May 2009 08:08:20 -0000http://mloss.org/software/rss/comments/187http://mloss.org/software/view/187/svmclassificationregressionmulti classlarge scale learningmultilabelrankingoptimizationbundle methodscutting plane