Implementation of English Handwritten Character Recognition System using Neural Network

Implementation of English Handwritten Character Recognition System using Neural Network

Madhav Sharma

Research Scholar (Ph.D.)

Computer science & Engineering

Faculty of Engineering & Technology

Jagannath University, Jaipur

Abstract: The technique of figuring out the handwritten textual content using device interface is known as handwritten man or woman popularity. Handwritten characters gift demanding situations due to the fact they range from one author to the other; even though the same author writes the identical character, the form, size, and place of the character vary. Handwritten character recognition has been one of the difficult areas in the field of picture processing. It has numerous applications which consist of, financial institution cheques and conversion of any handwritten report into digital text shape. There’s a outstanding want forstoring statistics to desktop garage from the facts to be had in handwritten images to later reprocess this facts utilizing computers. But to re-system this facts, it might be very hard to examine the other information from these picture files. Consequently, a way to automatically
clear up and save information, in particular text, from photograph documents is needed. This paper ambition to classify a character handwritten using dropout changes to 0.10 to 0.75 and implement the result as per epochs changes.  Model will trained and softmax technique is used to train model using modified gradient descent technique.

Keywords: Handwritten, Neural Network, Gradient Descent, Softmax, Trained