Novel Algorithm for Image Classification Using Cross Deep Learning Technique

Abstract

Abstract The basic concepts of training and the model structure of deep belief networks (DBNs) in deep analysis are studied to apply image recognition in the area of deep learning. Random propound is provided with the parameter in the fine-tuning stage and the randomly hidden layer eliminated to maintain unchanged weights. The results show that the layered DBN training system reduces training problems and training times significantly. In the small sample, the deep faith network has improved significantly after introducing the down sample and random dropdown and effectively alleviates the over-fitting phenomenon. Design a new Deep Learning Image Recognition and Classification Algorithm. Novel Algorithm for Image Classification Using Cross Deep Learning Technique.