For those of you not familiar with neural networks, they are a relatively new technology. A neural network is a mathematical or computational model that attempts to simulate the functionalities of the human brain by means of reverse engineering. The great thing about neural networks is their ability to learn. By specifying the parameters in which the network can learn and how it functions, we give it the guidelines it needs to learn from its mistakes and produce consistent, reliable results.
How are Neural Networks valuable?
There are a vast number of uses for neural networks, including but not limited to character recognition, time series prediction, pattern and sequence recognition, stock predictions and robotics. Any task where the complexity of the data and design is not viable to create in physical form, or when certain decisions are required to be based on experience or from lessons learnt through trial and error.
On the Web
Normally, when dealing in web environments, using such processor dependant systems as neural networks are not considered favourable. However, one thing I’ve learned in the web industry is that you should never limit imagination when it comes to web applications. As time progresses, server and client systems are only become more and more powerful. So why not push the boundaries of the web?
One Australian developer has done just that, with his network neural framework simply called ‘Neural Mesh’. Developed purely in PHP with a MySQL database he has used many optimisations to ensure the quickest processing and training of the network. The good news is that PHP and MySQL are readily available on most hosting servers around the world, so there’s little stopping us from delving into this new platform and producing some of the most advanced web applications to date.