Research on Improved Multi-label Deep Forest Algorithm
LI Cunhe
LI Shuai
Abstract：Multi-label Deep Forest(MLDF)is a deep ensemble learning model based on deep forest.In order to limit the com-plexity of the model and allow users to optimize the evaluation indexes according to their needs,two algorithms are used in multi-la-bel deep forest:measure-aware feature reuse and measure-aware layer growth.The former reuses the better features of the previous layer,while the latter is used to limit model growth.However,the measure-aware feature reuse algorithm uses the output of the cur-rent layer to take arithmetic average value to calculate the confidence,ignoring the impact caused by the accuracy difference of each forest.Therefore,an improved measure-aware feature reuse algorithm is proposed,which calculated the confidence by assigning weights to each label of each forest.Experimental results show that the improved algorithm has some improvement on low dimension-al multi-label datasets.
Keywords：Multi-Label Deep Forestmeasure-aware feature reuseconfidence
Publication Date：2023-12-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：5( 2790-2794 )