ABSTRACT
Deep learning (DL) has emerged as one of the most successful tools in image and signal processing. Excellent predictive performance in various high-profile applications has led to an enormous increase in focus on DL methods in a wide variety of fields, including artificial intelligence, astrophysics, chemistry and material science, healthcare, and manufacturing. The DL paradigm relies on multi-layer neural networks, which have been around for decades, with recent success due mostly to advances in computing power, algorithms, and the availability of enormous datasets for training. This has been propelled by the rise of graphical processing units (GPUs). Many dedicated toolboxes have been developed and continue to be developed, including Apache MXNet, Caffe, the Microsoft Cognitive Toolkit, PyTorch, and TensorFlow. Many thousands of papers are published every year on deep learning.
