• Ebooks
• In print books
• Used books
• Magazines
• Music
• Movie
• Credit card

 

Book Description: Since the first edition was published in 1997, some important advances have taken place in the theory of learning. * The notion of 'fat shattering' has been introduced. This theory permits an analysis of the learnability of real-valued functions, which is more applicable to practical situations than the older 'concept learning'. * The introduction of support vector machines (SVM's) has proved to be quite successful in some difficult applications, such as hand written character recognition. A theory of why SVM's work so well has now been developed. * The traditional assumption that learning takes place via a series of 'independent' observations has been removed through the use of so-called 'mixing processes'. The use of mixing process theory permits the learning paradigm to encompass the system identification problem, in which one tries to learn a 'dynamical system'. In contrast, traditional learning theory is applicable to static learning situations. * The idea of using randomised algorithms to solve difficult robust control problems has been introduced in the second edition, and has been effectively demonstrated in practical applications. The theory of learning control systems has now advanced to the level of learning neural networks.

[ For web hosting, AddALL recommend Liquidweb]