Transcript ChenPresentationQ3-0..

Applications of Neural Networks
By Jeff Chen
Background

What is a neuron?

types of neurons

the sigmoid function

how neurons are connected.
Background

What is a neural network?

types of neural networks

training methods

uses of neural networks
Backpropogation Network

Network by layers



Training



each layer has neurons that connect only to the layer
directly above and directly below it
each neurons has weights to its parents
training occurs by finding the error between the output
and the expected output
this is “propogated” back through the layers, each
neuron being assigned “blame” through its weights
Great for function approximation when the
function is not clear, is discontinuous, etc.
Competitive Network

Not just one network


Training





series of networks, all competing for the data
each network runs the input data
the network with the closest output to the desired
output “wins”
the “winnning” network then adjusts itself to better fit
the desired output
this way, each network begins to specialize in one
section or type of input
Great for classifying data when characteristics are
not obvious or are complicated.
Introduction

To create a Neural Network

Components of a Neuron




Input
Output
Sigmoid
Components of a Network



Neurons
Connections
Cases
Applications and Scope

Predict the stock
market



Train on previous days
Predict next day's
prices
Uses Backpropogation
network
Applications and Scope

Classify music



Takes in notes and
instruments
Classifies to time period of
music
Uses Competitive
network
Development




Using Java 1.5
File I/O
Network efficiency
Training on data sets
Results and Conclusions

Backpropogation network



Competitive network


Algorithm needs tweaking
Finds patterns very quickly
Converges too quickly
Larger purpose


A blackbox java implementation of neural networks
Flexible, can be adapted