Scribd
Upload
93 views11 pages

Problems On Som

This document describes the process of self-organizing maps for unsupervised learning classification. It shows input data being assigned to initial weight clusters. The weights are then updated through an iterative process as each input is evaluated to minimize distance to a weight cluster. This has the effect of the weights self-organizing to better represent the clustering of the input data.

Uploaded by

sd8837
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as XLSX, PDF, TXT or read online on Scribd
93 views11 pages

Problems On Som

This document describes the process of self-organizing maps for unsupervised learning classification. It shows input data being assigned to initial weight clusters. The weights are then updated through an iterative process as each input is evaluated to minimize distance to a weight cluster. This has the effect of the weights self-organizing to better represent the clustering of the input data.

Uploaded by

sd8837
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as XLSX, PDF, TXT or read online on Scribd
You are on page 1/ 11

Self Organizing Maps Unsupervised learning Classification

Input

x1 1 0 0 1
x2 0 1 1 0 w1 w2 clusters
x3 0 1 1 1
x4 1 0 1 1

x1 x2 x3 x4
Weights w1 w2
1 0.2 0.3
2 0.3 0.25
3 0.15 0.2
4 0.3 0.2

X1= [1,0,0,1]
D(1) 1.114675
D(2) 1.11018
D(2)<D(1) T=2 Update W2

Wj(new) = Wj(Old) + 0.5(X(i)- W(j)old))

w2new=w2old+0.5(xi-wjold)

Update W2
xi=[1001] w2old xi-w2old
W2new 0.3 1 0.3 0.7
0.25 + 0.5 0 0.25 -0.25
0.2 0- 0.2 -0.2
0.2 1 0.2 0.8
0.3 0.7 0.35
0.25 + 0.5 -0.25 -0.125 (o.5*(xi-wjold)
0.2 -0.2 -0.1
0.2 0.8 0.4

w2old (o.5*(xi-wjold)
0.3 0.35
0.25 + -0.125
0.2 -0.1
0.2 0.4

0.65
0.125
W2new 0.1 the input x1 is in w2
0.6

Weights w1 w2
1 0.2 0.65
2 0.3 0.125 epoch1(one sample is done)
3 0.15 0.1
4 0.3 0.6

X2=[0,1,1,0)
D(1) 1.158663
D(2) 1.535619
D(1)<D(2) T=1
Wj(new) = Wj(Old) + 0.5(X(i)- W(j)old))
W1new W1old+0.5(Xi-W1old)
x2 w1old x2-w1old
0.2 0 0.2 -0.2
0.3 1 0.3 0.7
0.15 + 0.5 1- 0.15 0.85
0.3 0 0.3 -0.3
x2-w1old
0.2 -0.2
0.3 0.7
0.15 + 0.5 0.85
0.3 -0.3

w1old 0.5*(x2-w1old)
0.2 -0.1
0.3 0.35
0.15 + 0.425
0.3 -0.15

0.1
0.65
W1new 0.575
0.15
x2 is in cluster 1
w1 w2
0.1 0.65 w1 w2
W 0.65 0.125 0.2 0.65
0.575 0.1 0.3 0.125
0.15 0.6 0.15 0.1
newly adjusted weights 0.3 0.6
previous weightd for the input x2
x3=[0111]
w1 w2
0.1 0.65
W 0.65 0.125
0.575 0.1
0.15 0.6
newly adjusted weights
d1 1.01891

d2 1.46906

d1<d2 w1 is to be updated

w1new=w1old+learningrate(xi-w1old)

0 0.1 -0.1
0.5 1 0.65 0.35
1 0.575 0.425
1 0.15 0.85

-0.1 -0.5
0.5 0.35 1.75
0.425 2.2
0.85 4,2

0.1 -0.5 -0.4 x3 is in cluster c1


0.65 1.75 2.4
0.575 2.2 2.775
0.15 4.2 4.35
ing Classification

2
c1 c2

x2 x1

x3

You might also like