Abstract:
This talk will explore the integration of learning modes into a single neural network structure in which layers of neurons and even individual neurons adopt different modes. There are several reasons to explore modal learning in neural networks. One motivation is to overcome the inherent limitations of any given mode (for example some modes memorise specific features, others average across features, and both approaches may be relevant according to the circumstances); another is inspiration from neuroscience, cognitive science and human learning, where it is impossible to build a serious model without consideration of multiple modes; and a third reason is non-stationary input data, or time-variant learning objectives, where the required mode is a funtion of time. Several modal learning ideas will be presented: The Snap-Drift Neural Network (SDNN) which toggles its learning between two modes, either unsupervised or guided by performance feedback (reinforcement); a general approach to swapping between several learning modes in real-time; and an adaptive function neural network (ADFUNN), in which adaptation applies simultaneously to both the weights and the individual neuron activation functions. Examples will be drawn from a range of applications such as natural language parsing, speech processing, geographical location systems, optical character recognition and virtual learning environments.

Brief Biography of the Speaker:
Dominic Palmer-Brown is professor of neural computing and Associate Head, School of Computing and Technology, at the University of East London. He was formerly chair in neurocomputing at Leeds Metropolitan University. His research covers neural network learning methods for processing language, modelling interaction and data mining. He was the neural network specialist on a 5 year UN/NERC/DoE funded crops data analysis project involving 15 countries, ending in 2000, and has supervised 12 phds to completion.

He was selected as Editor of the journal Trends in Cognitive Sciences by Elsevier Science London in 2001. He has published about 75 papers overall, and received best paper commendations at The International Conference on Education and Information Systems: Technologies and Applications (EISTA 2004) and the Int. Conf. on Hybrid Intelligent Systems 2003. He was keynote invited speaker at the European Simulation Multiconference 2003, and at The 10th Int. Conference on Engineering Applications of Neural Networks, 2007 and has published in many journals including IEEE Transactions in Neural Networks, Neurocomputing, and Connection Science.