| ||||
Mohamed Gaber: AbstractRandom Forests emerged as a highly accurate classifier and regressor over the last decade. Evidenced by both machine learning competitions (e.g., Kaggle’s) and scientific papers, Random Forests are considered to be in the forefront of the predictive machine learning techniques. In this talk, a number of adventurous developments to further improve the method will be discussed. These developments include ensemble pruning, feature engineering, and class decomposition for diversification. Achieved favourable experimental results will be presented. The talk will be concluded with pointers to future directions.
|