Self Organizing Maps (SOM

1
Self Organizing Maps (SOMs)

• Presented by
• Srilakshmi Gogula

2
Definition

• Self-organizing maps (SOMs) are a data
  visualization technique which reduce the
  dimensions of data through the use of
  self-organizing neural network (invented by
  Professor Teuvo Kohonen )
• Way SOMs follow to reduce dimensions - by
  producing a map of usually 1 or 2 dimensions
  which plot the similarities of the data by
  grouping similar data items together.

3
..contd

• SOM/SOFM is a type of artificial neural network
  that is trained using unsupervised learning.
• It operates in two modes
• -- Training (builds the map using input
  examples).
• -- Mapping (automatically classifies a new
  input vector)

4
Difference (SOMs and ANNs)

• Uses a neighborhood function to preserve the
  topological properties of the input space.
• This property makes SOM useful for visualizing
  low-dimensional views of high-dimensional data

5
SOM Example Figure
http//en.wikipedia.org/wiki/FileSynapse_Self-Org
anizing_Map.png
6
Main Algorithm

• Initialize the map
• Randomize the map's nodes weight vectors
• Grab an input vector
• Traverse each node in the map
• Use Euclidean distance formula to find similarity
  between the input vector and the map's nodes
  weight vector
• Track the node that produces the smallest
  distance (this node is BMU)

7
..contd

• Update the nodes in the neighborhood of BMU by
  pulling them closer to the input vector
• Wv(t 1) Wv(t) T(t)a(t)(D(t) - Wv(t))
• t current iteration
• ? limit on time iteration
• Wv current weight vector
• D target input
• T(t) restraint due to distance from BMU
• a(t) learning restraint due to time

8
..contd

• Increment t and repeat from 2 while t lt ?

9
Interpretation

• Training phase weights of the whole
  neighborhood are moved in the same direction
• More neurons point to regions with high training
  sample concentration and fewer where the samples
  are scarce.

10
Application Areas

• Image processing and speech recognition,
• Process control,
• Economical analysis
• Diagnostics in industry and in medicine.

11
Multiple Views

• It is a model of specific aspects of biological
  neural nets.
• It is a new paradigm in artificial intelligence
  and cognitive modeling.
• It is a tool for statistical analysis and
  visualization.
• It is a tool for the development of complex
  applications.

12
Advantages

• Very easy to understand
• Very simple
• -- Grey (Similar)
• -- Black ravine (Different)
• They work very well

13
Disadvantages

• Problem getting the right data
• They are computationally expensive
• Needs reconstruction in order to get one final
  good map

14
References

• http//davis.wpi.edu/matt/courses/soms/
• http//en.wikipedia.org/wiki/Self-organizing_map
• http//www.mlab.uiah.fi/timo/som/thesis-som.html
• http//www.english.ucsb.edu/grad/student-pages/jdo
  uglass/coursework/hyperliterature/soms/

15
Thank You