+ All Categories
Home > Documents > Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Date post: 29-Dec-2015
Category:
Upload: darcy-palmer
View: 218 times
Download: 1 times
Share this document with a friend
Popular Tags:
35
Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak
Transcript
Page 1: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Artificial Neural Network Theory and ApplicationAshish Venugopal

Sriram Gollapalli

Ulas Bardak

Page 2: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Presentation Overview

Artificial Neural Networks NASA profiles as a Pattern

Classification Task Implementing Pattern Classification

on the Silicon Recognition Neuron Board

Page 3: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

ANN General Overview

Inspired by biological neuron modelsDecision are distributed throughout

systemComponents interacts at several stagesComponents work in unison to solve

one problem ANN well suited to problems that don’t

have closed form algorithmic solution Black box model – hard to interpret

Page 4: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

ANNs Applied

Driving a car. Task involves determining the direction the road is curving based on an imageImage pixels are inputCurve angle is outputLearned non-linear relationship

between pixels and roads• hard to understand the model

Page 5: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Theory

Input layer, Hidden layer, Output LayerInput – one “neuron” for each pixelHidden – set of neurons to store what

was learned.Output layer – one neuron for each of 5

turn directions – ie straight, slight left, hard right..

But what is a neuron?

Page 6: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Neurons

Math terms: Binary decision makers English terms: take in some input, decide

whether is “high” or “low”, and pass this value on.

Similar to neurons in the brain propagating signals.

Page 7: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

How is the model learned?

Once you have training examplesPair (feature set, classification)Run the back propagation algorithms

Learns connection weights between neurons, as well as within neuron transfer weights.

Network topology changes affects learning behavior!

Page 8: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Unsupervised Networks

So far we have used labeled training examples to learn relationship between features and response variable.

What if we are just exploring patterns that exist in the features?

Unsupervised learning / clustering…

Page 9: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Unsupervised Goal

Find prototypical points within data set that can be used to approximate distribution of the whole data set.

Compression / Visualization Typical methods

Define a measure of similarity between data points

Select initial prototypes Move prototypes to best fit the data –

minimize error

Page 10: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Kohonen Net

Unsupervised technique

Each output neuron represents a prototype point, input layer is presented with data

Page 11: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Self Organizing Map

Each prototype is pulled towards that data, and brings its closest prototypes a little closer as well…

SOM Demo

Page 12: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

How is it relevant to our problem?

We will go through:Definition of what we are facingHow we can use neural netsHow we can improve on this method

Page 13: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

LIDAR in Space

Data from LITE projectSend a laser beam down from a

shuttle and get reflection readings.Data gathered for 53 hours in 1994.40GB of data

Page 14: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Input

Page 15: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Problem

There is too much data to send and processEach pixel is a number to be

transmitted.Each reading (one column) contains

3000 pixels. Need a way to transmit the

information in a more compact fashion.

Page 16: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Applying Neural Nets

Some readings are very similar to each other.

Define classes that will contain multiple readings.

Define a representative for the class which is close enough to all the class members.

Page 17: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Applying Neural Nets (cont.)

In order to get class Train Kohonen Nets on data for

specific number of classes.Once trained, just pass in a new

reading (3000 data points) and get the class id. Just transfer the class id.

Go from 3000x3000 numbers to be transferred to 3000 numbers

Page 18: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Example

Page 19: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Result – 8 classes

Page 20: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Result – 16 Classes

Page 21: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Result – 64 Classes

Page 22: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Result – 128 Classes

Page 23: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Result – 256 Classes

Page 24: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Problems

Takes a while to train 256 classes took 44 hours on a dual

P3 1GHz computer with 1.5GB of RAM.

Not optimalClassifying by the complete reading is

wasteful.• Need one number for each reading

Page 25: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Improving NN

sky

Page 26: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Improving NN (cont.)

Easier version:Look at blocks of 50x50 pixels instead

of just 1 pixel.Classify as sky, etc.Can preprocess using kNN, basic

filters to get rid of noise in data.Classes can be formed from

sequence of 50 pixel blocks.

Page 27: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Improving NN (cont.)

Harder version:Use variable sized mesh

• In parts with just the sky use a big block.• In detailed parts use smaller blocks.

Something to think about…

Page 28: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Muren Board

Silicon Recognition, Inc. 2 ZISC078 Chips (156 neurons in

parallel) 1 Mb of memory Up to 1 million recognitions/second

Page 29: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

ZISC

ZISC, the cornerstone of the Muren board, completely overcomes the three major limitations of software-based pattern recognition applications: ZISC fully handles non-linearity and

fuzziness ZISC is a fully massively parallel processing

system. Recognition speed is consistent, regardless of the number of stored patterns

ZISC does not use any algorithms for programming the solution, ZISC is taught

Page 30: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

RBF Space mapping

Page 31: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Example

Source picture

(circuit board)

Result (after processing)

(shows good and bad points)

Page 32: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Process

Learning Classifying Training Testing with new data

Page 33: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

NASA Data

Page 34: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Further investigation

Can we write directly to the board? Ability to preprocess data?

Page 35: Artificial Neural Network Theory and Application Ashish Venugopal Sriram Gollapalli Ulas Bardak.

Questions?


Recommended