1 neural networks - basics artificial neural networks - basics uwe lämmel business school institute...

26
1 Neural Networks - Basics Artificial Neural Networks - Basics Uwe Lämmel Business School Institute of Business Informatics www.wi.hs-wismar.de/ ~laemmel [email protected] wismar.de

Post on 19-Dec-2015

234 views

Category:

Documents


1 download

TRANSCRIPT

1 Neural Networks - Basics

ArtificialNeural Networks - Basics

Uwe Lämmel

Business SchoolInstitute of

Business Informatics

www.wi.hs-wismar.de/~laemmel

[email protected]

2 Neural Networks - Basics

Neural Networks

• Idea

• Artificial Neuron & Network

• Supervised Learning

• Unsupervised Learning

• Data Mining – other Techniques

3 Neural Networks - Basics

Artificial Neuron & Network

• Idea• An artificial Neuron• Neural Network• Example• Learning• Application

4 Neural Networks - Basics

Idea

• A human being learns by example “learning by doing”

– seeing(Perception), Walking, Speaking,…

• Can a machine do the same?

• A human being uses his brain. A brain consists of millions of single cells.A cell is connected with ten thousands of other cell.

• Is it possible to simulate a similar structure on a computer?

5 Neural Networks - Basics

Idea

Artificial Neural Network– Information processing similar to

processes in a mammal brain– heavy parallel systems, – able to learn– great number of simple cells

? Is it useful to copy nature ? – wheel, aeroplane, ...

6 Neural Networks - Basics

Idea

• we need:– software neurons– software connections between

neurons– software learning algorithms

An artificial neural network functions in a similar way a natural neural network does.

7 Neural Networks - Basics

A biological Neuron

• Dendrits: (Input) Getting other

activations• Axon: (Output ) forward the

activation (from 1mm up to 1m long)

• Synapse: transfer of activation:– to other cells, e.g. Dendrits of

other neurons– a cell has about 1.000 to 10.000

connections to other cells• Cell Nucleus: (processing)

evaluation of activation

cell and nucleus

Axon(Neurit)

Dendrits

Synapsis

8 Neural Networks - Basics

Natural vs. Artificial Neuron

cell and nucleus

Axon(Neurit)

Dendrits

Synapsis

j

jjii ownet

),( iiacti netfact

)( iouti actfo

w1i w2i wji...

oi

9 Neural Networks - Basics

Abstraction

• Dendrits: weighted (real number) connections• Axon: output: real number• Synapse: ---

(identity: output is directly forwarded)• Cell nucleus:

unit contains simple functionsinput = (many) real numbersprocessing = activation functionoutput = real number (~activation)

10 Neural Networks - Basics

An artificial Neuron

net : input from the networkw : weight of a connectionact : activationfact : activation function

: bias/thresholdfout : output function (mostly

ID)o : output

j

jjii ownet

),( iiacti netfact

)( iouti actfo

w1i w2i wji...

oi

11 Neural Networks - Basics

Exercise: AND/OR -LTU

• Built a „network“ that works like an AND-function

• Built an OR-network• Try to built an XOR-network

LTU – Linear Threshold Unit

12 Neural Networks - Basics

A simple switch

• Neuron = AND-function• Find parameters:

– Input neurons 1,2 : a1,a2 input pattern,

– weights of edges: w1, w2

– bias • Now evaluate output o !

a1=__ a2=__

o=__

net = o1w1+o2

w2a = 1, if net >= 0, otherwise

o = a

w1=__ w2=__

13 Neural Networks - Basics

Mathematics in a Cell

• Propagation function (neuron input) neti(t) = ojwj = w1i o1 + w2i o2 + ...

• Activationai(t) – Activation at time t

• Activation function fact :ai(t+1) = fact(ai(t), neti(t), i)

i – bias

• Output function fout :

oi = fout(ai)

14 Neural Networks - Basics

Activation

rea l

u n re str icted

[0 ,1 ][-1 ,+ 1 ]

In te rva l

co n tin uo us

0 ,1-1 ,+ 1

- ,+

b in ary

{ -1 ,0 ,+ 1}{ -10 0 ,... ,+ 10 0}

m u lti v a lue

d iscre te

A c tiv a tion

15 Neural Networks - Basics

Bias function

-1,0

-0,5

0,0

0,5

1,0

-4,0 -2,0 0,0 2,0 4,0

Identity

-4,0

-2,0

0,0

2,0

4,0

-4,0 -2,5 -1,0 0,5 2,0 3,5

Activation Functions activation functions

aresigmoid functions

16 Neural Networks - Basics

y = tanh(c·x)

-1,0

-0,5

0,5

1,0

-0,6 0,6 1,0

c=1c=2

c=3

-1,0

Activation Functions

y = 1/(1+exp(-c·x))

0,5

1,0

-1,0 0,0 1,0

c=1c=3

c=10

Logistic function:

activation functions

aresigmoid functions

17 Neural Networks - Basics

Structure of a network

• layers– input layer – input neurons– output layer – output neurons– hidden layer – hidden neurons

• An n-layer network has:– n layer of connections which can be

trained

18 Neural Networks - Basics

Definition: A Neural Network …

• … is characterized by – connections of many (a lot of) simple units

(neurons) and– units exchanging signals via these

connections

• … is a – coherent, directed graph which has– weighted edges and– each node (neuron, unit )

contains a value (activation).

19 Neural Networks - Basics

XOR-network

• weights are set by hand

• fact = 1, net > = 0, otherwise

= 1.0

input

output

weight matrix wij: i\j 1 2 3 4 1 0 0 1 1.2 2 0 0 1 1.2 3 0 0 0 -2 4 0 0 0 0

20 Neural Networks - Basics

XOR-Example

• standard propagation function: neti(t) = oj(t)wji

• activation function = bias function– ai = 1, if neti(t)> i e.g. =0.5

0, otherwise– output function = Identity: oj = aj

• use EXCEL and built the XOR-example network!

21 Neural Networks - Basics

Learning

Network

changing network parameters

evaluation network errorlearning

examples

22 Neural Networks - Basics

Learning - can be done by:

1. Modification of the weight of a connection – most frequently used

2. Deleting connections– can be done by (1): w=0 w0

3. Modification of the bias of a neuron– Can be done by (1) using an extra neuron

4. changing functions (activation, propagation, output function)

5. Building new cells (GNG)6. Building new connections7. Deleting cells

23 Neural Networks - Basics

Learning

• supervised learning– We know the results for certain input pattern:

teaching input teaching output

– Network error is used to adapt weights– Fast, but not natural

• reinforced learning– We know whether output is right or wrong– Information is used to adapt weights – Slower than supervised; natural

• unsupervised learning– Network has to learn by itself;– Slow , natural

24 Neural Networks - Basics

Applications

Pattern recognition (text, numbers, faces)

Checking the quality of a surface Control of autonomous vehicles Monitoring of credit card accounts Data Mining

25 Neural Networks - Basics

Applications

Speech recognition Control of artificial limbs classification of galaxies Product orders (Supermarket) Forecast of energy

consumption Stock value forecast

26 Neural Networks - Basics

Application - Summary

Classification Clustering Forecast Pattern recognition

Learning by examples, generalization Recognition of not known structures