All about Self Organizing Maps (SOMs)

Introduction

Here today in this article we have a new type of tool that is one of the most important tools recently, which is the Self-Organizing Map Method (SOM) is a new and powerful software tool for visualizing multidimensional data. It transforms complex and nonlinear statistical relationships between high-dimensional data into simple geometric relationships on a low-dimensional screen similar to other data mining methods, and it applies and integrates statistical procedures for noisy data modeling and processing. Since it thus compresses information while preserving the most important topological and metric relationships of the raw data elements on the screen, it can also be considered as a qualitative abstraction.

What is the (SOFM)?

Self-organizing maps or Kohenin maps are a type of artificial neural network introduced by Teuvo Kohonen in the 1980s. (paper link) SOM is trained using unsupervised learning, which is slightly different from other artificial neural networks, SOM does not learn by backpropagation with SGD, but rather uses competitive learning to adjust the weights in neurons.

SOM is a neural network algorithm that relies on unsupervised learning in a data-dependent manner. Unlike supervised learning methods, SOM can be used to aggregate data without knowing the membership of the input data class. So it can be used to discover the inherent features of the problem. SOM has been successfully applied in many engineering applications, covering, for example, areas such as pattern recognition, image analysis, process monitoring and control, and fault diagnosis, and SOM has also proven to be a valuable tool in data mining and knowledge discovery with applications in full text. and financial data analysis

What is a self-organizing map in machine learning?

Also in machine learning, a self-organizing map (SOM) or self-organizing feature map (SOFM) is an unsupervised method of machine learning that is used to produce a low-dimensional (usually two-dimensional) representation of a higher-dimensional data set while preserving the topological structure of the data we are working on.

What is a self-organizing map in an artificial neural network?

A self-organizing map is one of the most popular unsupervised learning artificial neural networks, where the system has no prior knowledge about the features or characteristics of the input data and the labels of the output data class. The network learns to form classes/groups of sample input patterns according to their similarities.

How do SOMs work?

• SOMs initialize weights for each node and choose a random vector from the training data.
• SOMs scan each node to find the weights that are most likely to be the input vector. The winning node is called the Best Matching Unit (BMU).
• SOMs discover the BMU neighborhood, and the number of neighbors decreases over time.
• SOMs assign a winning weight to the sample vector. The closer the node is to the BMU, the more its weight changes.
• The farther the neighbor is from the BMU, the less he learns. SOMs repeat the second step for N iteration.

Below, see a diagram of the input vector with different colors. This data is fed to the SOM, which then converts the data into 2D RGB values. Finally, it separates and categorizes the different colors.

In this article, work has been done to explain the technology that we talked about, work has been done to define it, and work has been done to mention how it works, and I hope that you will benefit from this technology.

Mohamed B Mahmoud. Data Scientist.