NANCY Seminar – Antonello Lobianco (BETA)

Title:  Introducing BetaML, the Machine Learning Toolkit of the BETA lab

Abstract:  The Beta Machine Learning Toolkit (BetaML) is a software package that hosts algorithms for many typical machine learning tasks, including regression, classification, clustering (soft/hard), dimensionality reduction (both linear and nonlinear), and missing data imputation.  In addition to the algorithms, BetaML provides a comprehensive range of utility functions for data processing (such as scaling, activation functions, and encoding), as well as sampling functions (like partitioning, cross-validation, and batching) and model performance measures (like confusion matrix, gini, and BIC/AIC).  Because algorithms are all self-contained in the library itself, the utility functions have APIs that coordinate with the algorithms, facilitating the preparation of the data for the analysis, the selection of the hyper-parameters and the evaluation of the models.  Most models have an interface for the MLJ framework and can be accessed natively in Julia.  Alternatively, they can be accessed in R or Python using `JuliaCall` and `PyJulia` respectively.  This presentation provides a general introduction to the library and showcases some examples of machine learning workflows supported by BetaML.