Binary Differential Evolution with Self-learning for Multi-objective Feature Selection
Description
This data "Binary Differential Evolution with Self-learning for Multi-objective Feature Selection" is a Matlab toolbox we developed for multi-objective feature selection based on an binary differential evolution with self-learning strategy, called MOFS-BDE. The details of this program can be found in the paper "Binary Differential Evolution with Self-learning for Multi-objective Feature Selection ", which has been submitted to the journal INFORMATION SCIENCES. 1. The idea of this algorithm Feature selection is an important preprocessing technique of data, which is generally modeled as a NP-hard multi-objective optimization problem. Focusing on Multi-Objective Feature Selection problems, this data shows our proposed multi-objective feature selection method, effective Binary Differential Evolution with self-learning strategy, called MOFS-BDE. First, a binary mutation based on probability difference is presented to locate potentially optimal areas fast by performing a large-scale search in the variable spaces. Second, a problem-specific self-learning strategy, named one-bit purifying search (OPS), is proposed to refine elite individuals obtained by the population in order to improve the exploitation performance of the algorithm. In addition, an efficient non-dominated sorting technology with the crowding distance is introduced to maintain the optimality and distribution of the population. 2. How to use this data or algorithm. (1) In this toolbox, the main function is named as “MOFSBDE”. In this function, you can select different data sets by changing the value of “fly”. (2) The file "data_t.m" saves all the test datasets; you can add new dataset you need; (3) the file "create.m" is used to update the population; (4) the file "non_domination_sort_mod1.m" is used to update the population by non-domination sorting method; (5) the file ""local_sech1.m" is our proposed the one-bit purifying search operator. (6) the final output is a matrix AC, which is a non-domination solution set. If you have any questions, please contact me in time. my email: YONGZH401@126.COM.
Files
Steps to reproduce
Step 1: open the main function “MOFSBDE” by Matlab; Step 2: select the test dataset by changing the value of “fly” in the main function; Step3: set the parameters of the proposed algorithm according your needs； Step 4: run the algorithm and output the final solution set "AC".