Neural network self-recommendation method based on greedy algorithm

The invention discloses a neural network self-recommendation method based on a greedy algorithm. The network is intelligently selected in three stages; in the first stage of network selection, 1/3D groups of test data are selected; a network model is selected according to the exploration probability Pe of 0.45; in the second stage of network selection, network models are selected according to theexploration probabilities Pe of 0.1 and 0.05; in the third stage of network selection, network models are selected according to the exploration probabilities Pe of 0.01 and 0.005 respectively, networkselection stopping conditions are set, and the network model with the detection precision, the detection precision ranking, the Mean 30 fluctuation ratio and the detection speed meeting the requirements serves as the optimal network model. According to the invention, the optimal network model can be obtained only by about 1/2D iterative operation; the accurate ranking of all network models in theneural network model set i