Parallel Approaches to Accelerate Deep Learning Processes Using Heterogeneous Computing

In the current context, the rise of artificial intelligence (AI) emphasizes the need to expedite training procedures, especially when dealing with extensive data, particularly in deep learning. This research primarily aims to significantly improve the time efficiency of deep learning processes. While it's widely recognized that graphics processing units (GPUs) offer notably faster performance for specific data tasks compared to a computer's central processing unit (CPU), this study explores heterogeneous computing systems for situations where GPUs are unavailable. Here, we investigate strategies to achieve enhanced processing speed using advanced technologies. The study concludes by presenting comparative results from various approaches and providing important recommendations for future endeavors.