High-Performance Computing and Its Requirements in Deep Learning

Modern-day computing involves the computation of complex, multi-dimensional and volumetric data-the generation of these data, basically from the web world, networking, and high-end applications. Natural images, medical images, hyperspectral images, and video datasets are always in complex form. To handle these data and run the application that generates these kinds of data and especially solve the complex problems related to these datasets, high-performance computing is the ultimate choice. The recent success and trending of deep neural networks for solving computer vision applications is a big leap in the computation direction. So, to solve complex computer vision problems with the deep neural network, high-performance computing is again the good choice, as deep neural networks are with deep architecture and many more trainable parameters. In this chapter, we cover many more aspects of high-performance computing and its requirements in deep learning in a competitive approach to help the potential researchers furthers.