IIP06-NN-Con 智能，信息系统，fuzzy，NN，neural，.pdfVIP

Page: 1 of 42 Bachelor of Science Bachelor of Science MIIS02 Intelligent Information Processing Basic Concepts of Neural Networks Dr. Liya Ding Liya/MIT/IIP06-NN-Con.ppt (2011) Page: 2 of 42 Outline Outline • Basic concepts of NN learning » Single layer Perceptron and linear separability » Multilayer Perceptron • General architecture of neural networks • Knowledge-based systems NN systems • Multilayer perceptron: » Need new learning algorithms Liya/MIT/IIP06-NN-Con.ppt (2011) Page: 3 of 42 Can Computer Really Think Learn Can Computer Really Think Learn • Thinking learning involve complex pattern recognition • In many real-world applications, computers are expected to perform complex pattern recognition. • Recognition of complex patterns consisting of numerous elements needs parallel processing parallel processing » Human can easily handle » Conventional computers are not suited to this type of problem Liya/MIT/IIP06-NN-Con.ppt (2011) Page: 4 of 42 Can Computer Really Think Learn (cont.) Can Computer Really Think Learn • Computers can do things being exactly told by programming • Computers can “inference” for decision making with complete rule set. » it is necessary to “tell”computers the relationship between a given situation and a conc