Neural Networks in 3 Minutes
By ai-depot | June 30, 2002
A clear introduction to neural networks, concise and to the point. Covers biological neurons to mathematical models, applications and motivations as well as future research. Dedicated to those of you that only have so much time on their hands… now there’s no excuse!
Written by Alex J. Champandard.
From Biology to Computing
This is a brief overview of Artificial Neural Networks in plain english, which will take about 3 minutes of your time. So lets get straight to business!
Neurons
Human and animal brains are large collections of inter-connected neurons. Neurons are complex cells that communicate together in elaborate patterns.
The dendritic tree collects stimuli from surrounding neurons, causing the body to react. In some cases it sends an electrical impulse down the axon. This spike will then in turn stimulate dentrites of neighbouring neurons.
Our attempts to understand how the human brain works includes artificially simulating it, from whence came the need for a simple mathematical model of a neuron.
These neurons need to be connected together to do something useful.
Topologies
There are various possible layouts for a specific number of neurons, this is called the topology of the network. Some are more ‘realistic’ representations of biological neural networks, and some have advantages for solving specific problems in computer science.
Types
- Feed forward - Simple, no backwards connections (shown top right)
- Recurrent - More complex behaviour, feed-back allowed (shown bottom right)
Varieties
- Sparse - Only some neurons have links to others (shown top right)
- Fully connected - All neurons are interconnected (shown bottom right)
Tags: none
Category: tutorial |
