An engine for high performance multi-agent environments with very large numbers of agents, along with a set of reference environments
created at Sept. 5, 2022, 4:29 p.m.
Support vector machines (SVMs) and related kernel-based learning algorithms are a well-known class of machine learning algorithms, for non-parametric classification and regression. liquidSVM is an implementation of SVMs whose key features are: fully integrated hyper-parameter selection, extreme speed on both small and large data sets, full flexibility for experts, and inclusion of a variety of different learning scenarios: multi-class classification, ROC, and Neyman-Pearson learning, and least-squares, quantile, and expectile regression.
created at April 22, 2017, 10:16 p.m.
ThunderGBM: Fast GBDTs and Random Forests on GPUs
created at Nov. 11, 2016, 9:58 a.m.
LibXtract is a simple, portable, lightweight library of audio feature extraction functions.
created at May 31, 2012, 3:16 p.m.
ThunderSVM: A Fast SVM Library on GPUs and CPUs
created at Dec. 11, 2014, 4:24 a.m.