Machine Learning Model Training Acceleration and Optimization Methods for Large-scale Datasets
Xiao Li, Hongcai Feng, Tianheng Pan, Xiaoling Wu · 2024
As the scale of data continues to grow, the training of machine learning models presents new challenges. In this paper, a machine learning model is established, which is optimized by stochastic gradient descent algorithm to solve the defects of traditional algorithm. The distributed training framework and distributed training system are proposed, which reduce the communication load effectively and improve the overall performance of the system. In this paper, a distributed training cluster consisting of 8 computers is used to verify the training acceleration and optimization methods of machine learning models on the CIFAR-10 dataset. The results show that compared with the other two methods, the training time of our method is the shortest, which is 1/h, the model accuracy is the highest, and the performance is always above 0.90. It can be seen that the research method in this paper can effectively improve the training efficiency and model accuracy, and has instructive significance.