Digit Recognition Using Machine Learning

unknownlippsAI and Robotics

Oct 16, 2013 (3 years and 5 months ago)

59 views

Digit Recognition Using

Machine Learning

Matheus Lelis

University of Massachusetts: Dartmouth

Abstract


The goal of this research project is to use an artificial
neural network machine learning algorithms with back
propagation to develop a program which will recognize
handwritten letters and numbers.

Background

Artificial Neural Network




An artificial neural network
learning algorithm is a learning
algorithm that is inspired by the
structure and functional aspects of
biological neural networks.




Computations are structured in
terms of an interconnected group
of artificial neurons, processing
information using a connectionist
approach to computation.

Problem

Character Recognition



Reading in the images of
handwritten numbers and
letters and outputting the
machine
-
encoded version.



Adding distortion to try to
solve CAPTCHAs.


Progress

The machine was written
using MATLAB works in
two steps.

1
st

step


runs through a set
of data and learns and sets
weight

2
nd

step


runs through new
data and tries to guess.

It takes in a matrix of data
with images 20px by 20px

Issues


Finding/Creating new data to test the machine with.



Teaching the machine to work with letters, only works
with numbers for now.



Adding the distortion and test out the CAPTCHAs.