Cloete I., Zurada J.M. (eds.) Knowledge-Based Neurocomputing

MIT Press, 1999. — 499.Knowledge-based neurocomputing concerns the utilization of problem-specific knowledge within the neurocomputing paradigm. The novelty of this book is that it illustrates the use of explicit comprehensible knowledge, a feature not always available within artificial neural networks. We believe that the issue of explicit knowledge manipulation within the neurocomputing paradigm has matured to such an extent, and is of such importance, that it warrants a volume dedicated to the review and exposition of approaches addressing this issue. This book is an outgrowth of that belief.
Uses of knowledge include prior knowledge about the problem domain, and the extraction, refinement and revision of knowledge about a specific problem domain contained within a neurocomputing system. The book also gives a thorough introduction to this field, describes the state-of-the-art methods, and points out the emergent directions for future research.
The book will be useful not only to those in the neurocomputing community, who wish to know more about the integration of knowledge-based principles and artificial neural networks, but also to the Artificial Intelligence community with its rich history of symbolic knowledge representation formalisms. The architectural issues addressed will aid developers aiming at connectionist-symbolic integration and the use of knowledge in hybrid intelligent systems.
The first part of the book (chapters 1 to 5) presents a taxonomy and reviews of methods for obtaining comprehensible neural network topologies by learning, as well as the encoding, extraction and refinement of symbolic knowledge within a neural network. The latter part presents wide ranging examples of applications of this technology, including rule extraction for chemical engineering systems, encoding and parameter estimation for differential equations describing known properties of a dynamical system, control systems development, time series analysis and prediction, and a neural expert system.
This book will therefore be valuable to researchers, graduate students and interested laymen in the areas of engineering, computer science, artificial intelligence, machine learning and neurocomputing, who require a comprehensive introduction to the subject, as well as keeping up with the latest research and the issues relevant to this exciting field.Knowledge-Based Neurocomputing: Past, Present, and Future
Architectures and Techniques for Knowledge-Based Neurocomputing
Symbolic Knowledge Representat ion in Recurrent Neural Networks: Insights from Theoretical Models of Computation
A Tutorial on Neurocornputing of Structures
Structural Learning and Rule Discovery
VLIANN: Transformation of Rules to Artificial Neural Networks
Integration of Heterogeneous Sources of Partial Domain Knowledge
Approximation of Differential Equations Using Neural Networks
Fynesse: A Hybrid Architecture for Self-Learning Control
Data Mining Techniques for Designing Neural Network Time Series Predictors
Extraction of Decision Trees from Artificial Neural Networks
Extraction of Linguistic Rules from Data via Neural Networks and Fuzzy Approximation
Neural Knowledge Processing in Expert Systems