Open Access. Powered by Scholars. Published by Universities.®
Physical Sciences and Mathematics Commons™
Open Access. Powered by Scholars. Published by Universities.®
Articles 1 - 5 of 5
Full-Text Articles in Physical Sciences and Mathematics
A Provably Convergent Dynamic Training Method For Multi-Layer Perceptron Networks, Timothy L. Andersen, Tony R. Martinez
A Provably Convergent Dynamic Training Method For Multi-Layer Perceptron Networks, Timothy L. Andersen, Tony R. Martinez
Faculty Publications
This paper presents a new method for training multi-layer perceptron networks called DMP1 (Dynamic Multilayer Perceptron 1). The method is based upon a divide and conquer approach which builds networks in the form of binary trees, dynamically allocating nodes and layers as needed. The individual nodes of the network are trained using a genetic algorithm. The method is capable of handling real-valued inputs and a proof is given concerning its convergence properties of the basic model. Simulation results show that DMP1 performs favorably in comparison with other learning algorithms.
An Integrated Framework For Learning And Reasoning, Christophe G. Giraud-Carrier, Tony R. Martinez
An Integrated Framework For Learning And Reasoning, Christophe G. Giraud-Carrier, Tony R. Martinez
Faculty Publications
Learning and reasoning are both aspects of what is considered to be intelligence. Their studies within AI have been separated historically, learning being the topic of machine learning and neural networks, and reasoning falling under classical (or symbolic) AI. However, learning and reasoning are in many ways interdependent. This paper discusses the nature of some of these interdependencies and proposes a general framework called FLARE, that combines inductive learning using prior knowledge together with reasoning in a propositional setting. Several examples that test the framework are presented, including classical induction, many important reasoning protocols and two simple expert systems.
Surface Intersection Loop Destruction, Thomas W. Sederberg, Alan K. Zundel
Surface Intersection Loop Destruction, Thomas W. Sederberg, Alan K. Zundel
Faculty Publications
The intersection curve between two surface patches consists of one or more connected components or branches. Each component can be classified as either an open branch, with endpoints on at least one patch boundary, or as a closed loop.
Hodographs And Normals Of Rational Curves And Surfaces, Thomas W. Sederberg, Takafumi Saito, Guo-Jin Wang
Hodographs And Normals Of Rational Curves And Surfaces, Thomas W. Sederberg, Takafumi Saito, Guo-Jin Wang
Faculty Publications
Derivatives and normals of rational Bézier curves and surface patches are discussed. A non-uniformly scaled hodograph of a degree m x n tensor-product rational surface, which provides correct derivative direction but not magnitude, can be written as a degree (2m - 2) x 2n or 2m x (2n - 2) vector function in polynomial Bézier form. Likewise, the scaled normal direction is degree (3m - 2) x(3n - 2). Efficient methods are developed for bounding these directions and the derivative magnitude.
Using Multiple Statistical Prototypes To Classify Continuously Valued Data, Tony R. Martinez, Dan A. Ventura
Using Multiple Statistical Prototypes To Classify Continuously Valued Data, Tony R. Martinez, Dan A. Ventura
Faculty Publications
Multiple Statistical Prototypes (MSP) is a modification of a standard minimum distance classification scheme that generates muItiple prototypes per class using a modified greedy heuristic. Empirical comparison of MSP with other well-known learning algorithms shows MSP to be a robust algorithm that uses a very simple premise to produce good generalization and achieve parsimonious hypothesis representation.