基于神经网络的交互动画解算优化方案

In order to improve the efficiency of subspace neural network model for physical understanding of computer interactive animation, this paper writes through data compression, structural pruning, and parallel computing instruction sets from three aspects: data preprocessing, neural network model compression, and front-end reasoning acceleration. As well as the method of adjusting the accuracy of the inference model, an optimization scheme for the existing algorithm is proposed. This scheme can effectively reduce the redundant data in the interactive animation, improve the effectiveness of the training set and speed up the training process of the neural network. At the same time, the scheme also It can effectively improve the sparsity of the neural network model to reduce the amount of calculation, and accelerate the front-end reasoning process of the neural network model on different target platforms.