Code

  • Regularized Empirical Risk Minimizer

    A C++ implementation of several binary and multiclass smooth loss functions including logistic regression, t-logistic regression, mismatched loss, and savage loss for regularized risk minimization. It supports both L1 and L2 regularizer. The package is implemented with optimized efficiency which features parallel optimization fueled by PETSc and TAO packages. [.tar.gz]

    (More completed documentation under construction)

  • t-logistic Regression

    A Matlab implementation of the t-logistic regression algorithm for binary classification. [.zip]

  • Nonparametric Bayesian Matrix Factorization by Power EP

    A Matlab implementation of the Infinite Matrix Factorization (IMF) and Infinite Nonnegative Matrix Factorization (INMF) algorithms. [.zip]