Parallel-FST: A feature selection library for multicore clusters

Feature selection is a subfield of machine learning focused on reducing the dimensionality datasets by performing computationally intensive process. This work presents Parallel-FST, publicly available parallel library for feature that includes seven methods which follow hybrid MPI/multithreaded approach to reduce their runtime when executed high performance computing systems. Performance tests were carried out 256-core cluster, where Parallel-FST obtained speedups up 229x representative and it was able analyze 512 GB dataset, not previously possible with sequential counterpart due memory constraints.