Year: 2004
Type: article
Abstract: Hybrid Intelligent Systems (HIS) combine intelligent techniques in synergistic architectures in order to provide solutions for complex problems.These systems utilize at least two of the three techniques: fuzzy logic, genetic algorithms and neural networks.The goal of their combination is to amplify their strengths and complement their weaknesses.Research in hybrid intelligent systems primarily focuses on the integration and interaction of different techniques rather than merging different methods to create new techniques.However, it is not always obvious or easy to build HIS architectures that achieve the higher intelligence goal.A good architecture for a hybrid system should match each of its tasks to the appropriate intelligent technique and provide an efficient means for their integration.Classification of signal patterns represents a complex problem due to the voluminous nature of signal patterns.A signal pattern is the combination of a large sequence of values of one or more variables collected over a period of time.Noise is an intrinsic component to all signal pattern applications.Current classification methods are inadequate in classifying large sets of noisy signal patterns.We selected this problem as the target for our new HIS architecture.In this research we developed a new HIS architecture for classifying large sets of signal patterns.Our Hybrid Intelligent Signal Pattern Classifier (HISPC) has demonstrated superior performance with low classification cost and great flexibility on synthetic and real life large data sets.An equally important objective of this research is the study of the software engineering aspect for developing this architecture.A new software engineering process and a set of design and implementation principles were developed in this work.These tools are applicable to the development of any experimental software system.
Authors: Roshdy S. Youssif, Carla Purdy
Institution: University of Cincinnati
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2256072054