id	author	title	date	pages	extension	mime	words	sentence	flesch	summary	cache	txt
bracis-28361	Silva, Sammuel Ramos; Silva, Rodrigo	FeatGeNN: Improving Model Performance for Tabular Data with Correlation-Based Feature Extraction	2023		.htm	text/html	5551	273	47	The proposed method consists of four phases: Pre-processing to reduce dimensionality, where the authors perform feature selection based on information gain (IG); Mining of correlated features to define and search for pairwise correlated features, where the distance correlation [8] is calculated to determine if there is an interesting predictive relationship between a pair of features; Feature generation, where regularized regression algorithms are used to search for associations between features and generate new features; and Feature selection, where features that do not add new information to the dataset are discarded. To address these challenges, we propose a novel convolutional method called FeatGeNN that extracts and creates new features using correlation as a pooling function.	cache/bracis-28361.htm	txt/bracis-28361.txt