GSPSO-LRF-ELM: Grid Search and Particle Swarm Optimization-Based Local Receptive Field-Enabled Extreme Learning Machine for Surface Defects Detection and Classification on the Magnetic Tiles

Algorithm 2

: PSO optimization for the initial weight A^init.

(1)	Particle swarm algorithm initialization begins:
(2)	generating D initial particles;
(3)	Calculating a feature map matrix, a pool map matrix, and an output weight
(4)	matrix β corresponding to the particles;
(5)	The prediction label T is obtained by the formula Hβ = T;
(6)	Obtaining a fitness function;
(7)	The update operation begins:
(8)	Individual update obtains the best value of the individual;
(9)	Global update obtains the best value of the global;
(10)	The optimal value is the optimal initial weight A^init;