An artificial neural network is a name for a kind of computer technology that tries to imitate the human brain. An artificial neural network or ANN includes simulated neurons and stimuli for attempts at reproducing the functions of the brain. This broad range of software and devices uses neural algorithm models to create decision making processes that planners hope will closely mimic human thought processes. Artificial neural networks represent a great advancement from relatively primitive ideas about computers in previous decades.
Neural network software is traditionally applied to game playing and other tasks that involve relatively calculated human thought. In a more bio-physical sense, neural networks are based on examination of how the brain’s neurons communicate and relay messages. Neural network applications include the interaction of various functions, where engineers look at the total productive output to see how these artificial neural network systems can effectively imitate human thought. A variety of “real-life applications” for ANN include regression analysis, function approximation, robotics, and general data processing.
Various types of artificial neural networks have been developed for different research provisions. These use different kinds of learning models such as supervised, unsupervised, or reinforced learning. Types of neural networks include a one-way feedforward neural network, a radial basis function or RBF network, a Kohonen self organizing network, and even modular neural networks where a larger network is made up of several small ones.
Another type of new structure applied to artificial neural networks is often called a “committee of machines” where various network structures each provide their own “vote” or “opinion” in a decision modeling process. This is also sometimes called an associative neural network or ASNN. The benefit of this kind of research is evident to engineers who believe that ASNN can help model human group decision making or other complex modeling in some similar ways to the individual decision making models provided by ANN.
A principle that is often utilized by an artificial neural network is called “fuzzy logic.” The word “fuzzy” is used to describe any gaps in data or knowledge. Neural networks are often able to close some data or knowledge gaps by educated guessing and statistical prediction, which is in contrast the strict "yes or no" binary logic traditionally associated with electronic decision making. Overcoming fuzzy logic helps neural networks to provide better results in simulations. Using the building blocks of previous research, planners and engineers experienced with artificial neural networks are continually enhancing what these tools can do for pushing the boundaries of our knowledge about our own minds.