Applied Intelligent Systems: New Directions

ISBN-13: 9783540211532 / Angielski / Twarda / 2004 / 348 str.

John Fulcher; Lakhmi C. Jain

Opis

Humans have always been hopeless at predicting the future...most people now generally agree that the margin of viability in prophecy appears to be 1 ten years. Even sophisticated research endeavours in this arena tend to go 2 off the rails after a decade or so. The computer industry has been particularly prone to bold (and often way off the mark) predictions, for example: 'I think there is a world market for maybe five computers' Thomas J. Watson, IBM Chairman (1943), 'I have traveled the length and breadth of this country and talked with the best people, and I can assure you that data processing is a fad that won't last out the year' Prentice Hall Editor (1957), 'There is no reason why anyone would want a computer in their home' Ken Olsen, founder of DEC (1977) and '640K ought to be enough for anybody' Bill Gates, CEO Microsoft (1981). 3 The field of Artificial Intelligence - right from its inception - has been particularly plagued by 'bold prediction syndrome', and often by leading practitioners who should know better. AI has received a lot of bad press 4 over the decades, and a lot of it deservedly so. How often have we groaned in despair at the latest 'by the year-20xx, we will all have...(insert your own particular 'hobby horse' here - e. g.

Spis treści

1 Adaptive Technical Analysis in the Financial Markets Using Machine Learning: a Statistical View.- 1.1 ‘Technical Analysis’ in Finance: a Brief Background.- 1.2 The ‘Moving Windows’ Paradigm.- 1.3 Post-Hoc Performance Assessment.- 1.3.1 The Effect of Dividends.- 1.3.2 Transaction Costs Approximations.- 1.4 Genetic programming.- 1.5 Support-Vector Machines.- 1.6 Neural Networks.- 1.7 Discussion.- References.- 2 Higher Order Neural Networks for Satellite Weather Prediction.- 2.1 Introduction.- 2.2 Higher Order Neural Networks.- 2.2.1 Polynomial Higher-Order Neural Networks.- 2.2.2 Trigonometric Higher-Order Neural Networks.- Output Neurons in THONN Model#1.- Second Hidden Layer Neurons in THONN Model#1.- First Hidden Layer Neurons in THONN Model#1.- 2.2.3 Neuron-Adaptive Higher-Order Neural Network.- 2.3 Artificial Neural Network Groups.- 2.3.1 ANN Groups.- 2.3.2 PHONN, THONN & NAHONN Groups.- 2.4 Weather Forecasting & ANNs.- 2.5 HONN Models for Half-hour Rainfall Prediction.- 2.5.1 PT-HONN Model.- 2.5.2 A-PHONN Model.- 2.5.3 M-PHONN Model.- 2.5.4 Satellite Rainfall Estimation Results.- 2.6 ANSER System for Rainfall Estimation.- 2.6.1 ANSER Architecture.- 2.6.2 ANSER Operation.- 2.6.3 Reasoning Network Based on ANN Groups.- 2.6.4 Rainfall Estimation Results.- 2.7 Summary.- Acknowledgements.- References.- Appendix-A Second Hidden Layer (multiply) Neurons.- Appendix-B First Hidden Layer Neurons.- 3 Independent Component Analysis.- 3.1 Introduction.- 3.2 Independent Component Analysis Methods.- 3.2.1 Basic Principles and Background.- 3.2.2 Mutual Information Methods.- 3.2.3 InfoMax ICA Algorithm.- 3.2.4 Natural/Relative Gradient Methods.- 3.2.5 Extended InfoMax.- 3.2.6 Adaptive Mutual Information.- 3.2.7 Fixed Point ICA Algorithm.- 3.2.8 Decorrelation and Rotation Methods.- 3.2.9 Comon Decorrelation and Rotation Algorithm.- 3.2.10 Temporal Decorrelation Methods.- 3.2.11 Molgedey and Schuster Temporal Correlation Algorithm.- 3.2.12 Spatio-temporal ICA Methods.- 3.2.13 Cumulant Tensor Methods.- 3.2.14 Nonlinear Decorrelation Methods.- 3.3 Applications of ICA.- 3.3.1 Guidelines for Applications of ICA.- 3.3.2 Biomedical Signal Processing.- 3.3.3 Extracting Speech from Noise.- 3.3.4 Unsupervised Classification Using ICA.- 3.3.5 Computational Finance.- 3.4 Open Problems for ICA Research.- 3.5 Summary.- References.- Appendix — Selected ICA Resources.- 4 Regulatory Applications of Artificial Intelligence.- 4.1 Introduction.- 4.2 Solution Spaces, Data and Mining.- 4.3 Artificial Intelligence in Context.- 4.4 Anomaly Detection: ANNs for Prediction/Classification.- 4.4.1 Training to Classify on Spare Data Sets.- 4.4.2 Training to Predict on Dense Data Sets.- 4.4.3 Feature Selection for and Performance of Anomaly Detection Suites.- 4.4.4 Interpreting Anomalies.- 4.4.5 Other Approaches to Anomaly Detection.- 4.4.6 Variations of BackProp’ ANNs for Use with Complex Data Sets.- 4.5 Formulating Expert Systems to Identify Common Events of Interest.- A Note on the Software.- Acknowledgements.- References.- 5 An Introduction to Collective Intelligence.- 5.1 Collective Intelligence.- 5.1.1 A Simple Example of Stigmergy at Work.- 5.2 The Power of Collective Action.- 5.3 Optimisation.- 5.3.1 Optimisation in General.- 5.3.2 Shades of Optimisation.- 5.3.3 Exploitation versus Exploration.- 5.3.4 Example of Common Optimisation Problems.- Minimum Path Length.- Function Optimisation.- Sorting.- Multi-Component Optimisation.- 5.4 Ant Colony Optimisation.- 5.4.1 Ant Systems — the Basic Algorithm.- The Problem with AS.- 5.4.2 Ant Colony Systems.- 5.4.3 Ant Multi-Tour System (AMTS).- 5.4.4 Limiting the Pheromone Density — the Max-Min Ant System.- 5.4.5 An Example: Using Ants to Solve a (simple) TSP.- 5.4.6 Practical Considerations.- 5.4.7 Adding a Local Heuristic.- 5.4.8 Other Uses for ACO.- 5.4.9 Using Ants to Sort.- An Example of Sorting Using ACO.- 5.5 Particle Swarm Optimisation.- 5.5.1 The Basic Particle Swarm Optimisation Algorithm.- 5.5.2 Limitations of the Basic Algorithm.- 5.5.3 Modifications to the Basic PSO Algorithm.- Choosing the Position S.- The Problem of a finite t.- Aggressively Searching Swarms.- Adding Memory to Each Particle.- 5.5.4 Performance.- 5.5.5 Solving TSP Problems Using PSO.- PSO Performance on a TSP.- 5.5.6 Practical Considerations.- 5.5.7 Scalability and Adaptability.- References.- 6 Where are all the Mobile Robots?.- 6.1 Introduction.- 6.2 Commercial Applications.- 6.2.1 Robot Couriers.- 6.2.2 Robot Vacuum Cleaners.- 6.2.3 Robot Lawn Mowers.- 6.2.4 Robot Pool Cleaners.- 6.2.5 Robot People Transporter.- 6.2.6 Robot Toys.- 6.2.7 Other Applications.- 6.2.8 Getting a Robot to Market.- 6.2.9 Wheeled Mobile Robot Research.- 6.3 Research Directions.- 6.4 Conclusion.- A Note on the Figures.- References.- 7 Building Intelligent Legal Decision Support Systems: Past Practice and Future Challenges.- 7.1 Introduction.- 7.1.1 Benefits of Legal Decision Support Systems to the Legal Profession.- 7.1.2 Current Research in AI and Law.- 7.2 Jurisprudential Principles for Developing Intelligent Legal Knowledge-Based Systems.- 7.2.1 Reasoning with Open Texture.- 7.2.2 The Inadequacies of Modelling Law as a Series of Rules.- 7.2.3 Landmark and Commonplace Cases.- 7.3 Early Legal Decision Support Systems.- 7.3.1 Rule-Based Reasoning.- 7.3.2 Case-Based Reasoning and Hybrid Systems.- 7.3.3 Knowledge Discovery in Legal Databases.- 7.3.4 Evaluation of Legal Knowledge-Based Systems.- 7.3.5 Explanation and Argumentation in Legal Knowledge-Based Systems.- 7.4 Legal Decision Support on the World Wide Web.- 7.4.1 Legal Knowledge on the WWW.- 7.4.2 Legal Ontologies.- 7.4.3 Negotiation Support Systems.- 7.5 Conclusion.- Acknowledgements.- References.- 8 Forming Human-Agent Teams within Hostile Environments.- 8.1 Introduction.- 8.2 Background.- 8.3 Cognitive Engineering.- 8.4 Research Challenge.- 8.4.1 Human-Agent Teaming.- 8.4.2 Agent Learning.- 8.5 The Research Environment.- 8.5.1 The Concept of Situational Awareness.- 8.5.2 The Unreal Tournament Game Platform.- 8.5.3 The Jack Agent.- 8.6 The Research Application.- 8.6.1 The Human Agent Team.- 8.6.2 The Simulated World Within Unreal Tournament.- 8.6.3 Interacting With Unreal Tournament.- 8.6.4 The Java Extension.- 8.6.5 The Jack Component.- 8.7 Demonstration System.- 8.7.1 Wrapping Behaviours in Capabilities.- 8.7.2 The Exploring Behaviours.- 8.7.3 The Defending Behaviour.- 8.8 Conclusions.- Acknowledgements.- References.- 9 Fuzzy Multivariate Auto-Regression Method and its Application.- 9.1 Introduction.- 9.2 Fuzzy Data Analysis.- 9.2.1 Fuzzy Regression.- 9.2.2 Fuzzy Time Series Analysis.- 9.2.3 Fuzzy Linear Regression (FLR).- Basic Definitions.- Linear Programming Problem.- 9.3 Fuzzy Multivariate Auto-Regression Algorithm.- Example — Gas Furnace Data Processed by MAR.- 9.3.1 Model Selection.- 9.3.2 Motivation for FLR in Fuzzy MAR.- 9.3.3 Fuzzification of Multivariate Auto-Regression.- 9.3.4 Bayesian Information Criterion in Fuzzy MAR.- 9.3.5 Obtaining a Linear Function for a Variable.- 9.3.6 Processing of Multivariate Data.- 9.4 Experimental Results.- 9.4.1 Experiments with Gas Furnace Data.- 9.4.2 Experiments with Interest Rate Data.- 9.4.3 Discussion of Experimental Results.- 9.5 Conclusions.- References.- 10 Selective Attention Adaptive Resonance theory and Object Recognition.- 10.1 Introduction.- 10.2 Adaptive Resonance Theory (ART).- 10.2.1 Limitations of ART’s Attentional Subsystem with Cluttered Inputs.- 10.3 Selective Attention Adaptive Resonance Theory.- 10.3.1 Neural Network Implementation of SAART.- Postsynaptic Cellular Activity.- Excitatory Postsynaptic Potential.- Lateral Competition.- Transmitter Dynamics.- 10.3.2 Translation-invariant 2D Shape Recognition.- 10.4 Conclusions.- References.

Blurb

This carefully edited book presents examples of the successful application of Intelligent Systems techniques to practical problems. The invited contributions, written by international experts in their respective fields, clearly demonstrate what can be achieved when AI systems are used to solve real-world problems. The book covers the field of applied intelligent systems with a broad and deep selection of topics, such as object recognition, robotics, satellite weather prediction, or economics with an industrial focus. This book will be of interest to researchers interested in applied intelligent systems/AI, as well as to engineers and programmers in industry.