The Missouri University of Science and Technology hosts the largest research program in Adaptive Critic Designs (ACDs), a neural-network-based generalization of dynamic programming for optimization and control. Other topics in computational intelligence, and particularly in neural networks, are also well-represented. The research groups in the ECE, CS, EMGT, MAEEM and other Departments are constantly securing new funding and publishing new results. The ACIL is located in the ECE Department, with many collaborations. Our results include several firsts among ACD implementations and applications, and theoretical results generalizing the last quarter century of research in the field. The lab is active in many areas of neural network theory and application. Other neural net designs used include Adaptive Resonance Theory, Probabilistic Neural Networks, backpropagation including backpropagation through time, recurrent multilayer perceptrons, Support Vector Machines, Node-Decoupled Extended Kalman Filter training, and other popular as well as custom designs. Applications include military logistics, aerospace and automotive control problems, bioreactor control, various control benchmarks, cotton quality analysis, crop modeling, tornado prediction, intelligent agents, the game of Go, and financial forecasting. Theoretical contributions include generalization of approximate dynamic programming architectures, analysis of machine learning approaches for intelligent agents, explanation capabilities of neural networks, and scaling properties. Applications include knowledge discovery in databases and Internet search techniques. The ACIL also is active in fuzzy logic. Theoretical results include improvements to the fuzzy extension principle and better techniques for representing fuzzy weights in neural networks. Both of these remove possible inconsistencies from the results of fuzzy computations. Applications include qualitative risk assessment, intelligent agents, and control.
Students working in the laboratory gain many advantages, including collaboration in a work environment, continued involvement with research, the positive influence of role models and mentors, and, more often than not, an opportunity to publish. (Publishing is required for all graduate students.) The ACIL welcomes small and large business cooperative ventures in intelligent computing.
Research projects for business have included...
- Intelligent Control Systems
- Integrated Vehicle Health Management
- Path planning in smart sensor networks
- Telecommunications networking
- VLSI hardware
- Optical neural networks
Financial forecasting and engineering.
We've also compiled a list of other Neural Network Resources on the Internet and will be continuing to update it!