Introduction to Neural NetworkNeural NetworkWhat is an Artificial Neural Network? (ANN)Single NeuronThree Layers Neural NetControl Systems ApplicationThe areas where neural nets may be usefulLimits to Neural NetworksThe Advantage Using Neural NetworkThree Main ApplicationsExample – Pattern RecognitionQuestions?Introduction to Neural Network Justin JansenDecember 9th 2002Neural Network•Definition of Artificial Neural Network•Fundamental concepts•Where they fit in a control system•What they can do and where they fail•Example pattern recognitionWhat is an Artificial Neural Network? (ANN)•A neural network is a computational method inspired by studies of the brain and nervous systems in biological organisms. •A common neural network architecture consists of multiple layers of similar elements: –Each unit is called a neuron and is capable of receiving input stimulation). –When the total amount of stimulation received exceeds some predetermined threshold, the neuron "fires" –When highly interconnected, produce dynamic response to inputsStructure of a neuron in a neural net Single NeuronNeural net with three neuron layers Three Layers Neural NetControl Systems Application· Pattern association · Pattern classification · Regularity detection · Image processing · Speech analysis · Optimization problems · Robot steering · Processing of inaccurate or incomplete inputs · Stock market forecasting · Simulation The areas where neural nets may be useful•The operational problem encountered when attempting to simulate the parallelism of neural networks •Instability to explain any results that they obtain •Neural networks are, in essence, a "black box" •Training timeLimits to Neural Networks•Handle partial lack of system understanding •Create adaptive models (models that can learn)•Noted for their ability to learn patterns in data which is noisy •Excellent for situations in which the trainer is unsure of the actual relationships that exist in the training set. The Advantage Using Neural Network•Concurrent simulation, Neural Network results have been validated by the expected real-life behavior. •Neural Networks as sub-components of a bigger model to model subsystems that would be hard to model commonly because of a lack of understanding. •Adaptive models, "models that can learn", according to an error feedback such model would be able to adapt runtime to situations that hasn't been taken into account.Three Main ApplicationsExample – Pattern Recognition•Neural Network Best Fit
View Full Document
Unlocking...