METHOD AND SYSTEM FOR HYPER-PARAMETER OPTIMIZATION AND FEATURE TUNING OF MACHINE LEARNING ALGORITHMS

摘要

The present disclosure provides a method for hyper-parameter optimization and feature tuning of machine learning algorithms. The method comprises: A: randomly generating an initial collection pool including a plurality of parameter ensembles; B: performing an EnKF-based iterative optimization on the plurality of parameter ensembles in the initial collection pool respectively to obtain a plurality of first optimized parameter ensembles; C: evaluating the plurality of first optimized parameter ensembles respectively to obtain a first collection pool and a first supplement parameter collection pool; D: updating the initial collection pool with the first supplement parameter collection pool; repeating steps B, C and D a first predetermined number of times to obtain a second collection pool and a set of optimal parameters. The method increases calculation efficiency and accuracy of the hyper-parameter optimization and feature tuning, and has strong universality. Besides, the present disclosure further provides a system for hyper-parameter optimization and feature tuning of machine learning algorithms.