Recent advances in machine learning (ML), fueled by new frameworks and algorithms, more powerful computing architectures, scalable cloud-based services, and availability of large-scale data sets, have enabled scientists and engineers to tackle more complex problems than ever before. Computer hardware has made tremendous leaps in processing power, bit depth, caching, and storage. Graphics processing units, developed originally for the gaming industry, provide a parallel processing capability that is ideally suited to computer vision (CV) and the simulation of artificial neural networks. The expansion to cloud-based services allows researchers virtually unlimited scaling of resources to tackle problems having millions of input attributes, with output domains up to thousands of potentially nonexclusive classes.
For more about this article see link below.