Engineering and design

Cybernetics and Neural Networks

Module code: 100H6
Level 7 (Masters)
15 credits in autumn semester
Teaching method: Laboratory, Lecture
Assessment modes: Unseen examination, Coursework

A cybernetic device responds and adapts to a changing environment in a sensible way. Neural network systems permit the construction of such devices exploiting information, feedback and control to achieve intelligent interaction and behaviour from autonomous devices such as robots.

In this module the utilisation of artificial intelligence techniques and neural networks are explored in detail. Software implementation of theoretical concepts will solve genuine engineering problems in dynamic feedback control systems, pattern recognition and scheduling problems. In many instances solutions must be computed in response to data arriving in real-time (e.g. video data). The implications of high speed decision making will be explored.

The module will explore:

  • neuron models
  • network architectures
  • perceptron and perceptron learning rule
  • synaptic vector spaces
  • linear transformations for neural networks
  • supervised hebbian learning
  • performance optimisation
  • widrow-hoff learning
  • associative learning
  • competitive networks.

Learning will be supported by laboratories using the Matlab Neural Network Toolbox.

Module learning outcomes

  • The fundamental principles of neural network systems and their applications.
  • A range of specialist topics related to neural network systems.
  • Current problems and emerging solutions in the applications of neural networks.
  • The analytical and practical techniques applicable to advanced scholarship in neural networks systems.