Deep learning is a:
“Field of study that gives computers the ability to learn without being explicitly programmed”
Arthur Samuel, 1959
Machine learning, and specifically deep learning, has been and continues to be an extremely disruptive technology in many fields of computer science. The success of deep learning techniques is based on solving extremely difficult selective grouping and regression problems which has resulted in their prompting acceptance by academics for solving complex problems.
The emergence of deep learning is generally accredited to a cycle of mastery and understanding of complexity whereby fundamental developments in training deeper models were assisted by the accessibility to enormous data sets and high-performance computer hardware.
Deep learning algorithms continue to evolve at a speedy pace. Early on, frameworks were subjugated to the availability of matrix multiplication libraries. However, more finely tuned algorithms have been developed over decades as research has continued with the introduction of newer and more sophisticated kinds of algorithms.
The ensuing enlightenment was the need to go beyond generic matrix multiplication and therefore, complex networks came along which resulted in even more innovative algorithms. Many of these algorithms are designed manually using assembly language. Nevertheless, low-level tweaks can lead to amazing performance enhancements. For some operations such as group stability performance can increase many times over when compared to non-optimized solution.
Here are some examples of resources for Academic and non-academic applications:
School of Engineering and Applied Sciences Harvard University
Tutorial introduction to deep learning architecture and networks
MICRO Tutorial (2016) MIT
http://eyeriss.mit.edu/tutorial.... Joel Emer, Vivienne Sze, Yu-Hsin Chen
Google’s Tensor Flow models are coarse-grained dataflow graphs Basic building block is an “operation” https://www.tensorflow.org/insta...
No comments:
Post a Comment