Neural Network Tutorials

• Introduction to neural networks
  
• Structure of a neural network
  
• Types of neural networks
  
• Neural network training
  
• BackPropagation Networks
  
• The Learning Rate
  
• Backpropagation Algorithm Enhancements
  
• Neural Network Applications
  

Introduction to neural networks

Artificial Neural Network is an information processing paradigm that attempts to simulate the functionality of human brain.

"Why do we need neural networks? What is the need to simulate human brain when we have a computer which can perform millions of numerical calculations within a fraction of a second?" The reason is, even though a computer can store huge amounts of data and can perform numerical calculations efficiently, there are some problems which a computer struggles to solve.

Let us take an example, we have text based search engines that are capable of searching for a keyword in the internet. Thinking in similar lines, imagine an image searching application, where the user is given some space to draw an image and the application should search all image files similar to the image drawn. If the user draws a tree, the search engine should return all images having a tree. Why don't we have such systems available? Ofcourse, there are search engines like 'Google Image Search' which are meant to search image files, but in fact they only search for a matching filename. Searching for an image using text!! Can't we search images directly?

This image search belongs to a broad class of similar problems, called Pattern Recognition problems. Character recognition, hand-writing recognition, face recognition, speech recognition, image processing, signature recognition, speaker recognition and face recognition are some of the challenges that belong to this pattern recognition class.

Though a computer can store huge amounts of data, it does not know how to recognize and interpret the data. On the other hand, human brain is incredibly efficient in recognizing patterns. Just by looking at one's face, we can recognize who the person is (if we know him), but it is not so easy for a computer. This is because, there is no particular algorithm which can map a picture (or a video) to a name. It is not possible to mathematically model pattern recognition. This makes it impossible to solve such problems using conventional problem solving techniques. If we want to solve pattern recognition, we have to somehow simulate human brain. Artficial neural networks is an attempt to do the same.