Accelerating CNN training on the MNIST dataset Using Parallel Computing Technologies

2025/01/10 by Rakhimov, Mekhriddin, Javliev, Shakhzod

paper · doi:10.34920/icdgpdt.alkhws40

Abstract

The computational speed of computers is constantly rising due to the quick development of technology. However, the growing field of artificial intelligence necessitates more and more time and resources to implement machine learning processes. In this situation, using parallel processing methods on computers turns out to be a useful way to improve training process efficiency. The difficulties with computational speed in machine learning operations are discussed in this paper. It also examines the outcomes of employing Convolutional Neural Networks (CNN) to speed up the MNIST dataset’s training process.

Related

• Optimizing CNN Using HPC Tools
• Distributed learning of CNNs on heterogeneous CPU/GPU architectures
• Improving Strong-Scaling of CNN Training by Exploiting Finer-Grained Parallelism
• Implementation of Training Convolutional Neural Networks
• Recent Advances in Convolutional Neural Network Acceleration