Binary Differential Evolution with Self-learning for Multi-objective Feature Selection

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.