Wrap Up
Recent studies in the field of machine learning and its industrial applications have shown that ensemble machine learning methods are more accurate than individual algorithms due to their greater robustness to overfitting on many tasks. Ensemble machine learning methods include both homogeneous approaches, which use sets of identical algorithms, and heterogeneous approaches, which exploit greater diversity by combining different algorithms.
In this talk, different approaches used for ensemble machine learning in the context of financial decision support were compared. Particular attention was paid to demonstrations of state-of-the-art ensemble machine learning methods on classification and regression problems in finance, such as the detection of fraudulent financial transactions, credit risk modeling, financial distress prediction, product backorder predictions, and bitcoin price prediction.
