Synopsis

The module will provide an introduction to the field of artificial neural networks and and will survey specific connectionist computational models addressing cognitive/perceptual phenomena. We will study the development of perceptrons, assciative memories, supervised and unsupervised learning techniques, and recurrent networks. In the second part of the course cognitive models in the areas of development, vision, memory and language will be studied and compared.

We will work mainly from O'Reilly and Munakata (2000). For the first part of the course a more mathematical text is Hertz, Krogh and Palmer(1991), and a less mathematical text is the book by Kevin Gurney (1997).   For the second part of the course models will be drawn from O'Reilly and Munakata (2000)..

Lecture information

Module organizer, lecturer: Nigel Goddard room C19, 5 Forrest Hill, tel 650-3087
Email: Nigel.Goddard@ed.ac.uk
Times: ??
(First lecture Mon 8th October)
Location: ??

Tutorials

Students will be assigned to a tutorial group. Each group will meet once a week in parallel with the lectures. The tutors for this course are Nigel Goddard and Nicola de Pisapia (pisapia@anc).

Weekly lecture plan:

This rough schedule will evolve as the term progresses:

1. Introduction to Connectionist Cognitive Modeling.
   Hopfield Networks.
2. Perceptrons.
   Back propagation (possibly 2 lectures).
3. Unsupervised learning.
4. Recurrent networks, reinforcement learning.
5. Hippocampal memory. Place-cells versus other theories.
6. Development of the visual system. Topographic development, ocular dominance, orientation.
7. Face perception.
   Sequence learning.
8. Language acquisition.
9. Issues of representation.
   Future directions / Conclusions.

Assesment

Assessment will be by 1.5 hour sit-down closed book exam in the third term, worth 50% of the overall mark; several computer-based assignments due during the module, worth 50% of the mark.

References

O'Reilly, R.C., and Munakata, Y (2000), Computational Explorations in Cognitive Neuroscience: Understanding the Mind by Simulating the Brain, MIT Press.

Hertz, J., Krogh, A., Palmer, R.G. (1991), Introduction to the Theory of Neural Computation, Addison-Wesley Publishing.

Gurney, L (1997) An Introduction to Neural Networks, UCL Press. A draft of this book is available on-line.