Training of Dynamic Neural and Fuzzy-Neural Networks for Modeling and Control of Nonlinear Dynamic Systems


Level: Intermediate / Advanced | Free to ICCAR 2017 Delegates

Dynamic neural networks are a special type of networks having feedback connections which allow them to exhibit a dynamic behavior for processing sequential and time-varying patterns. They have been applied to solve diverse real-world problems involving temporal and dynamic characteristics. The tutorial presents the designing and training of dynamic neural networks: Back Propagation Through Time BPTT and Dynamic Back Propagation DBP algorithms are derived and used to train dynamic neural networks in supervised or reinforcement learning schemes. Dynamic neural networks are used for the modeling and control of dynamic systems. Static and dynamic feedback controllers are trained considering the internal dynamics of the system, as well as fuzzy-neural networks are designed considering human knowledge and experience. The concept of incremental learning is applied for assuring the successful training of neural networks from simple to complex tasks. Neural networks are applied for the autonomous control of car-like and trailer-type mobile robots.

Dr. Antonio Moran Cardenas

Dr. Antonio Moran obtained the Doctor and Master degrees in Mechanical Systems Engineering from Tokyo University of Agriculture and Technology, Japan, where he has been associate professor and scientific researcher in the Laboratory of Robotics and Control Systems. He has been president of the IEEE Robotics and Automation Society RAS, Peru Chapter, and obtained the 2014 Best Society Award in the International Conference in Robotics and Automation ICRA held in Hong Kong, China.
Dr. Moran is a visiting professor at Tokyo University of Agriculture and Technology, Japan, and Stockholm University, Sweden. He is a professor at the Graduate School of Pontifical Catholic University of Peru, and technical manager of Technova SAC, company providing engineering solutions to industry.
His research interests include computational intelligence, integration of neural networks, fuzzy logic and genetic algorithms, learning systems, mobile robots, nonlinear systems modeling and control, and their industry applications.