Research and design of distributed training algorithm for neural networks

This paper presents a new distributed training algorithm for neural networks based on multi-agents, which is created to solve the bottleneck of memory in current prediction of protein secondary structure program, and a chip training algorithm is proposed to work in the distributed environment to evolve the global optimum by competition from a group of neural network agents that have different sample chips for processing. The experimental results demonstrate that this method can effectively improve the convergent speed, has good expansibility, and can be applied to the prediction of protein secondary structure of middle and large size of amino-acid sequence.