Business Systems and Knowledge Modelling Lab

Research Project

Humanization of Soft Computing Agents

Research Goal
Most business applications employ soft computing technologies like neural networks, fuzzy logic, genetic algorithms and a range of other statistical techniques for prediction, forecasting, classification, clustering and decision making. However, all these technologies provide imprecise and approximate solutions. These soft computing solutions , firstly, need to be optimized in order to improve quality of their solution and performance on an ongoing basis. The optimisation can be done based on human feedback or other feedback from the environment. Secondly, they need to be modeled in the problem solving and decision making context of the managers to be effectively used.

This research aims to research, develop and apply multi-layered soft computing agent architecture (addressing the above two research issues) for business problems.

Funding
45000

Demonstrations
A prototype has been developed

Team Members
Rajiv Khosla, Chris Lai, Qiubang Li, Tharam Dillon

Publications
[1] R. Khosla, N. Ichalkaranje, L. Jain Design of Multi-agent systems : Human-Centredness, Architectures, Learning and Adaptation, Springer-Verlag, 2004

[2] R. Khosla, “Multi-layered Distributed Agent Ontology for Soft Computing Systems.” to appear in International Journal Knowledge-Based Intelligent Engineering Systems, 2004.

[3] R. Khosla and Q. Li, Unified Problem Modelling Language for Knowledge
Engineering of Complex Systems – Computational Level, to appear International
Journal of Soft Computing, Springer-Verlag, 2004

[4] R. Khosla, Component Based Distributed Multi-Agent Architecture for Soft
Computing in Soft Computing Agents: A New Perspective for Dynamic Information Systems, pp. 57-78. IOS Press.

[4] R. Khosla and T. Dillon, 'Task Structure Level Symbolic-Connectionist
Architecture', a chapter in a book titled Neuro-Symbolic Integration, edited by Ron
Sun and Frederic Alexandre, Lawrence Erlbaum Associates, USA, pp. 37-56,
November 1997.

[5] R.Khosla, “Multi-Layered Distributed Problem Solving Agent Ontology for Soft
Computing Systems,” to appear in the 2nd IEEE workshop on Advanced Environmental Sensing and Monitoring Technologies , IEEE Computer Society Como, Italy, July 2003.

[6] S. Yoshimori, Y. Mitsukura, M. Fukumi N. Akamatsu and R. Khosla, “License
Plate Detection in Rainy Days,” to appear in IEEE 2003 International Symposium on Computational Intelligence for Robotics and Automation, IEEE Computer Society, Kobe, Japan, July 2003

[7] Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu, Rajiv
Khosla, “An EEG Feature Detection System Using the Neural Networks Based on
Genetic Algorithms,” to appear in IEEE 2003 International Symposium on Computational Intelligence for Robotics and Automation, Kobe, IEEE Computer Society , Japan, July 2003

[8] R. Khosla, “Multi-Layered Distributed Agent Ontology for Soft Computing Systems,” to appear in the Seventh International Conference on Knowledge-based Intelligent Information and Engineering Systems, Oxford, U.K., pp. 445-52, September 2003

[9] J. Main, T. Dillon, and R. Khosla, 'Use of Neural Networks for Case-Retrieval in
a System for Fashion Shoe Design', in Eighth International Conference on Industrial
and Engineering Applications of AI and Expert Systems (IEA/AIE-95), June, 1995,
Melbourne, pp.151-8.

[10] J. Main, T. Dillon, and R. Khosla, 'Use of fuzzy feature vectors and neural
networks for case retrieval in case-based systems'. 1996 Biennial Conference of the
North American Fuzzy Information Processing Society (NAFIPS). University of
Berkley, California, USA, June 1996, pp. 438-43.

[11] R. Khosla and T. Dillon, 'Task Structure Level and Computational: Architectural
Issues in Symbolic-Connectionist Integration', International Joint Conference of Artificial Intelligence (IJCAI95), Working Notes of the IJCAI Workshop on Neuro- Symbolic Integration , Montreal, Canada, August, 1995, 119-25

[12] R. Khosla and T. Dillon , 'Constructs for Building Complex Symbolic- Connectionist Systems', in Sixth IEEE conference on Tools with Artificial Intelligence , New Orleans, Louisiana, U.S.A., November, 1994, pp. 93-6.

[13] S. K. Tang, T. Dillon, R. Khosla, 'Fuzzy Logic and Knowledge Representation in
a Symbolic-Subsymbolic Architecture', in 1995, IEEE International Conference on
Neural Networks, Perth, Australia, pp. 349-53.

[14] S. K. Tang, T. Dillon, and R. Khosla, 'Application of an Integrated Fuzzy,
Knowledge-Based, Connectionist Architecture for Fault Diagnosis in Power Systems', in Sixth Intelligent Systems Applications Conference, Florida, 1996, pp.
188-93.

[15] R. Khosla and T. Dillon, 'Cognitive and Computational Foundations of Symbolic-
Connectionist Integration' in ECAI'94 Workshop Proceedings on Symbolic- Connectionist Processing, Netherlands, August, 1994. pp.10-19.

[16] R. Khosla and T. Dillon, 'An Integrated Information Processing and Learning
Architecture' in Neuro-Nimes, Paris, France, 1994

[17] R. Khosla and T. Dillon , 'Perspectives for Integration of Artificial Neural Networks and Expert Systems', in IEEE Australia-New Zealand Conference on Information Systems , Perth, November, 1993, pp. 624-28

[18] R. Khosla and T. Dillon, 'Task Decomposition and Competing Expert System-
Artificial Neural Net Objects for Reliable and Real Time Inference', in International
IEEE Conference on Neural Networks, San Francisco, U.S.A, pp794-800, Mar/Apr.
1993.

[19] R. Khosla and T. Dillon, ' Intelligent Hybrid Multi-Agent Task Level Architecture
for Knowledge Engineering of Complex Systems', in IEEE International Conference on Neural Networks, Houston, Texas, USA, June 1997, vol. 3, pp. 1865-1872.

[20] R. Khosla, ' Human-Centered Virtual Machine of Problem Solving Agents
Intelligent Agents, Software Agents and Objects', invited paper in Third IEEE High
Assurance Systems Symposium (HASE’98), November, 1998

[21] R. Khosla and T. Dillon, Welding Symbolic AI Systems and Neural Networks for
Intelligent Systems, invited paper in IEEE International Conference on Neural Networks, Anchorage, Alaska, pp. 29-34, May 1998.

[22] R. Khosla, ' Intelligent Hybrid Multi-Agent Computational Level Architecture for
Knowledge Engineering of Complex Systems', in IEEE International Conference on Systems, Man and Cybernetics, Orlando, Florida, USA, October 1997, pp. 1412-19

[23] R. Khosla and T. Dillon, 'Fusion of knowledge-based systems and neural networks and applications'. Keynote Address. First IEEE International Conference on Conventional and Knowledge-Based Intelligent Electronic Systems (KES '97). Adelaide, Australia, May 21-23 1997, pp. 27-44

[24] R. Khosla, 'Task Structure Level and Computational Level, and Program Level:
Architectural Issues in Symbolic-Connectionist Integration', invited talk at the Neural Information Processing Systems (NIPS95) workshop on Symbolic Dynamics in Neural Processing Vail, Colorado, USA, November 1995.