Hybridization and optimization of machine learning techniques for improved forecasting in real-world scenarios

Abstract

Different and powerful machine learning paradigms are constantly in a race for delivering the lowest error and/or the highest comprehensibility. But what can certainly lead to better forecasting is model inter-cooperation or intra-optimization. The aim of the current talk is to put forward some recent ideas for such hybridization and optimization. Demonstrative experiments are outlined for problems coming from real, challenging environments.