 
Linton Miller's Abstracts
Evaluating Complex Systems using the Theory of Signal Detectability and
Information Theory
Abstract
Evaluating the performance of a computer system is an important part of its
development. Evaluation provides an objective way of assessing the system,
making comparisons with other systems, and making comparisons with theoretical
performance. In this seminar, I will examine current methods of evaluation and
their limitations, and present a new approach developed by Brian Scurfield
(Psychology Department) which addresses these limitations. The approach models
the computer system as a detection system, and then uses a new measure of
detectability, D, for evaluation. D is derived from
information theory and the Theory of Signal Detectability (TSD). I will give an
example by applying this new method to a patternrecognition neural network.
References
 Abramson, N. (1963) Information theory and coding. McGrawHill, New
York.
 Egan, J. P. (1975) Signal Detection Theory and ROC Analysis.
Academic Press, New York.
 Green, D. M. & Swets, J. A. (1966) Signal Detection Theory and
Psychophysics. Wiley, New York. Reprinted by Krieger, Huntingdon, NY,
1974.
 Scurfield, B. K. (1995) Discrimination
among events by neural networks. Ph. D. thesis, Victoria University of
Wellington, New Zealand.
 Scurfield, B. K. (1996). Multipleevent forcedchoice tasks in the theory
of signal detectability. Journal of Mathematical Psychology, 40 (3)
253269.
 Scurfield, B. K. (1998). Generalization of the theory of signal
detectability to mdimensional nevent forcedchoice tasks.
Journal of Mathematical Psychology, 42 (1) 531.

Shannon, C. (1949) The mathematical theory of communication.
In Shannon, C. & Weaver, W., The mathematical theory of communication,
pp. 399. The University of Illinois Press, Urbana.
Last updated
08 Nov 2009 04:37 PM
