A CNN Vehicle Recognition Algorithm based on Reinforcement Learning Error and Error-prone Samples

Abstract

Focused on the issue that in the latter part of training the recognition accuracy of convolutional neural network increased slowly and the training process was time-consuming, a reinforcement learning error training samples and error prone training samples CNN recognition algorithm was proposed. The algorithm uses the advantage that the network error rate decreases rapidly in the initial epochs of training. In the initial stage of training, reduce the training epochs and after each training, error training samples and error prone training samples are updated to the training set. Apply this algorithm to vehicle recognition. The accuracy of the experiment was 98.33%, and the training time was shortened to 30.97% of the original algorithm. Experimental results show that in the process of training learning error samples and error prone samples repeatedly improves the recognition accuracy and reduces the training time of CNN.