The course introduces the theory and practice of neural computation. It
offers the principles of neurocomputing with artificial neural networks widely
used for addressing real-world problems such as classification, regression, pattern
recognition, data mining, time-series prediction, etc.. Two main topics are covered:
supervised and unsupervised learning. Supervised learning is studied using linear
perceptrons, and non-linear models such as probabilistic neural networks, multilayer
perceptrons, and radial-basis function networks. Unsupervised learning is studied using
Kohonen networks. Recurrent networks of the Hopfield type are briefly covered. There
are offered contemporary training techniques for all these neural networks. Knowledge
and tools for the specification, design, and practical implementation of neural
networks are also provided.
back to top
The objectives of the course in neural networks are:
to introduce the neural networks as means for computational learning;
back to top
On successful completion of this module, students will be able to:
Subject-related Knowledge
Subject-related Skills
Transferrable Skills
back to top
Haykin, Simon.
Neural Networks. A Comprehensive Foundation.,
Second Edition, Prentice-Hall, Inc., New Jersey, 1999.
New Edition:
Haykin, Simon.
Neural Networks and Learning Machines.,
Pearson Higher Education, 2009.
back to top
The students are advised to attend the lectures. Attendance to workshops is compulsory and will be monitored.
back to top
Excellent A = 70 - 100
Very Good B = 60 - 69
Good C = 50 - 59
Acceptable D = 40 - 49
Weak E = 35 - 39
Fail F = 0 - 34
back to top
One Written Examination: 80%
One Coursework Assignment: 20%
Coursework Assignment: Training a feed-forward network with the backpropagation algorithm (an individual neural network architecture with specific weight parameters will be given to every student, and this network has to be trained with two different input examples).
Coursework Marking Scheme- every student should present the following:
a) Calculations to process the first training example
a.i) computed outputs from each network node during the forward pass; [4x2.5 marks]
a.ii) computed errors (betas) at each network node during the backward pass; [4x4 marks]
a.iii) computed changes (deltas) for each network weight; [8x2 marks]
a.iv) updated all network weights. [8x1 marks]
total after processing the first example: 50 marks
b) Calculations to process the second example
b.i) computed outputs from each network node during the forward pass; [4x2.5 marks]
b.ii) computed errors (betas) at each network node during the backward pass; [4x4 marks]
b.iii) computed changes (deltas) for each network weight; [8x2 marks]
b.iv) updated all network weights. [8x1 marks]
total after processing the second training example: 50 marks
Full marks will be awarded for results rounded up to and including the fourth digit after the decimal point.
Feedback on the Coursework Assignment: One week after the deadline the marks on the submissions will be ready, and every student will receive the comments on his work written by the lecturer on the presented worksheet (during the office hours or at convenient times by appointment).
back to top
Handout Lecture Coursework: 27 October 2010.
Submit Lecture Coursework: 15 December 2010.
back to top
Friday 13:00 - 14:00 p.m.
Friday 14:00 - 15:00 p.m.
back to top
Bishop, C.M.
Neural Networks for Pattern Recognition,
Oxford University Press, Oxford, UK, 1995.
Hagan,M.T., Demuth,H.B. and Beale,M.H.
Neural Network Design,
PWS Publ. Co., Boston, MA, 1996.
Ham,F.M. Kostanic,I.
Principles of Neurocomputing for Science and Engineering,
McGraw-Hill Publ. Co., Boston, MA, 1996.
Nikolaev,N. and Iba,H.
Adaptive Learning of Polynomial Networks: Genetic Programming,
Backpropagation and Bayesian Methods,
Springer, New York, 2006.
Principe,J.C., Euliano,N.R. and Lefebvre,W.C.
Neural and Adaptive Systems: Fundamentals Through Simulations.
Wiley, New York, 2000.
Reed,R.D., and Marks,R.J, II
Neural Smithing: Supervised Learning in Feedforward Artificial Neural Networks,
The MIT Press, Cambridge, MA, 1999.
back to top
Neural Networks Education Repository
Netlab: Neural Network Software in Matlab