Layered Approximation for Deep Neural Networks

Abstract:

Artificial Intelligence has created immense hype in the last decade and the credit for the same goes to the groundbreaking breakthrough named "Deep Learning" or "Deep Neural Network". Although "Artificial Neural Network", the foundation of Deep Learning as a concept has been prevalent since 1958 but the actual implementation for solving real business use cases has only been possible over the last decade. 

Deep Neural Networks has demonstrated significant results in the fields of computer vision, speech recognition, and machine translation, and outperformed human brain in many instances. Artificial Neural Network, as it is inspired from human biological neural superstructure has few structural similarities but not possible in terms of how the human brain or biological neural network works because we have still limited information on its functioning. Nevertheless, Deep Neural Network paved the way for many possibilities and currently, it is the most promising technology that we have in the field of Artificial Intelligence. 

Chakraborty, Utpal, Layered Approximation for Deep Neural Networks (April 6, 2018). Telecom Business Review, 12 (1) 2019, 18-22, Available at SSRN: https://ssrn.com/abstract=3449430