Você está na página 1de 17

EEN-351 :ARTIFICIAL NEURAL NETWORKS

Course Google Group: ANN-2018@googlegroups.com


Course Objectives: This course will introduce a variety of topics
in the broad area of neural networks, giving students a foundation
for studying this growing area. The course also will attempt to
provide an idea of both the strengths and weaknesses of neural
networks to problem solving.
Define artificial neural network yourself after understanding
the example
Classifying human breast tissues into normal
and malignant (cancerous)
Pathologically characterized 56 tissue samples were obtained
immediately after surgery.
The tissue samples were irradiated with polarized light of
wavelength 488 nm from a 5W Ar ion laser. Polarized
fluorescence light was collected by using an analyzer. The
fluorescence spectra were recorded using a SPEX 1877
triplemate and a PMT
Data set- 56x190 (190 fluorescence intensities)
28 patients whose normal and malignant breast tissues.
We have 56 samples. We divide these into training and testing sets.
First take the training samples. We know out of these 28 samples
how many are normal. We will do some ‘computation’ so that when
we present a normal sample to this black box it will produce a ‘0’
and when we present a malignant sample it will produce a ‘1’ We
use all these 28 training samples to teach the black box. Now we
take a sample from the testing set. When we input a normal sample,
the output must be 0

190 One output


dimensional which varies
input between 0 and 1
Black box
(does computation)
Classification vs. Regression
• Classification:
– predicts categorical class labels (discrete)
– classifies data (constructs a model) based on the
training set and the values (class labels) in a
classifying attribute and uses it in classifying new
data
• Regression:
– models continuous-valued functions

Classification: identify the class. Basically identify a


discrete quantity
Regression: estimate continuous quantity (tomorrow’s
temperature, stock market index)
Examples of problems
Input Output
• age, weight, blood • Time on a 5-mile run
pressure (regression)
• age, weight, blood • Finish or not finish a 5-mile
pressure run (Classification)
• number of bedrooms, days
on the market , median • Selling price of the house
price of houses sold in the (regression)
area last year
Classification: find function that maps inputs into
classes: class = f(x)
Regression: find function that maps inputs into
values: value = f(x)
Inputs are also called features

Features are properties of items to be classified that will aid


in discriminating between classes.

Discrete features: gender, nationality, blood type


Continuous features: location in room, barometric pressure,
value of dollar

If you choose good features, a problem is easy and almost


any solution method works. If you choose badly, nothing
works
x2 1 x1  1 x2  1.5  0
An Example:
2 x1  x2  1.5  0

 x1 
1 1    1.5  0
t

 x2 
8
1
Sl. x1 x2 y
No
7 linearly separable
1 0.7 0.7 0 w

2 0.8 0.9 1 0 1 2
x1
3 0.8 0.25 0
4 1.2 0.8 1 x1
x1
5 0.6 0.4 0 1 1
6 1.3 0.5 1 y y
x2 1
1 =1.5
7 0.9 0.5 0 x2 -1.5
8 0.9 1.1 1 1

wt x  b  0 w  1 b  1.5 b  


1 0
x2

Decision Boundary

x1

x2
x1

’Any function can be approximated to arbitrary accuracy by a


network with two hidden layers’

Neural network mysticism:


A neural network may solve a practical problem, but it can be
difficult to understand how it solved it. For many problems the
hidden layer is not doing an obvious analysis. If you don’t know
what was done, it can be hard to improve it.
Strengths of a Neural Network
• Power: Model complex functions, nonlinearity built
into the network
• Ease of use:
– Learn by example
– Very little user domain-specific expertise needed
• Intuitively appealing: based on model of biology, will
it lead to genuinely intelligent computers/robots?

Neural networks cannot do anything that cannot be


done using traditional computing techniques, BUT
they can do some things which would otherwise be
very difficult.
Syllabus

• Single Layer Perceptron


• Multilayer Perceptron
• ADALINE
• The LMS Algorithm
• Backpropagation Learning
• Overfitting, Cross-Validation, and Early Stopping
• Principal Components Analysis (PCA)
• Evolutionary methods (GA& PSO) for optimization
•Radial Basis Function Network
•Extreme Learning Machines
•Unsupervised Neural Networks
- LVQ, SOM, ART ..etc
• Support Vector Machines
•Dynamics of neural networks
•Hopfield Networks and Brain State in a Box (BSB) networks
• Neuro- fuzzy approaches
Books
Neural Networks for pattern recognition- C. M. Bishop
Neural Network Design by Martin T. Hagan & Howard B. Demuth & Mark H.
Beale
Neural Networks- A comprehensive foundation- SIMON HAYKIN
Elements of Artificial Neural Networks- K. Mehrotra, C. K. Mohan, Ranka
Introduction to Artificial Neural Systems- J.M. Zurada
Fundamentals of Artificial Neural Networks- Mohamad H Hasssoun
Fundamentals of Neural Networks – Architecture, algorithms and applications-
Laurene Fausett
Neural Networks- Algorithms, Applications, and Programming Techniques- James A.
Freeman/ David M. Skapura
An Introduction to Neural Networks- James A. Anderson
Neural Networks- A classroom approach- Satish Kumar (TMH)
Grading Policy

Task Weight
Quizzes, Homework, Viva 12.5%
Course Project 12.5%
Mid-term 25%
Final Exam 50%
Total 100%
Iris Problem Type load fisheriris in MATLAB then type who

Length and width of sepal and petal for three Northern American
species of iris. Fisher’s Iris data set contains 3 classes of 50 instances
each, where each class refers to a type of iris plant

1. sepal length in cm Example


Class
2. sepal width in cm 5.1 3.5 1.4 0.2 0
Iris Setosa: class 0
3. petal length in cm 4.9 3.0 1.4 0.2 2
Iris Versicolour: class 1
4. petal width in cm 4.6 3.1 1.5 0.2 1
Iris Virginica: class 2

Você também pode gostar