To Improve Performance of Number Guesser Neural Network by using Tensorflow and Deep Learning

Authors

  • K. Pavankalyan
  • M. Nalini

Abstract

The human visual framework is one of the miracles of the world. In every half of the globe of our mind, people have an essential visual cortex, otherwise called V1, containing 140 million neurons, with several billions of associations between them. But then human vision includes V1, however, a whole arrangement of visual cortices - V2, V3, V4, and V5 - doing continuously increasingly complex picture handling. We convey in our minds a supercomputer, tuned by advancement more than a huge number of years, and sublimely adjusted to comprehend the visual world. Perceiving manually written digits isn't simple. Or maybe, we people are spectacularly, astoundingly great at understanding what our eyes show us. Yet, almost all that work is done unknowingly. Thus we don't typically acknowledge how extreme an issue our visual frameworks explain. Neural systems alternately approach the issue. The thought is to take countless written by hand digits, known as preparing models, and afterward build up a framework that can gain from those preparation models. At the end of the day, the neural system utilizes the guides to naturally derive rules for perceiving manually written digits. Moreover, by expanding the quantity of preparing models, the system can study penmanship, thus improve its precision. So while I've indicated only 100 preparing digits above, maybe we could assemble a superior penmanship recognizer by utilizing thousands or even millions or billions of preparing models.

Downloads

Published

2020-02-01

Issue

Section

Articles