A Scene Classification Method based on Neural Network and Computational Verb

Deep learning is often used in learning tasks for complex scene classification. However, the neural network established by deep learning has a long process from initial state to convergence. It may take more time for training the neural network for a network with more complex structure and higher recognition rate. This paper chooses a neural network learning method based on computational verbs, by selecting an appropriate computational verb rules, which dynamically changes the neural network learning step size to achieve the purpose of allowing the neural network to converge quickly. The experimental results show that the convergence speed of the neural network can be accelerated under the premise of ensuring the classification accuracy by adding computational verbs in the training process.