Chapter 16

Application case study—machine learning

Boris Ginsburg

Abstract

This chapter provides an application study of how CUDA and GPU computing helped to enable deep learning and revolutionize the field of machine learning. It starts by introducing the basic concepts of convolutional neural networks (CNN). It then shows the CNN code examples that have been accelerated with CUDA. The chapter concludes with an explanation of how the cuDNN library uses a matrix multiplication formulation of the convolution layer computation to improve the speed and utilization of the GPU.

Keywords

Convolutional neural network; machine learning; deep learning; matrix–matrix multiplication; forward propagation; gradient backpropagation; training; cuDNN

Get Programming Massively Parallel Processors, 3rd Edition now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.