Approximations to ideal-observer performance on signal-detection tasks

The ideal-observer performance, as measured by the area under the receiver' s operating characteristic curve, is computed for six examples of signal-de tection tasks, Exact values for this quantity, as well as approximations ba sed on the signal-to-noise ratio of the log likelihood and the likelihood-g enerating function, are found. The noise models considered are normal, expo nential, Poisson, and two-sided exponential. The signal may affect the mean or the variance in each case. It is found that the approximation from the likelihood-generating function tracks well with the exact area, whereas the log-likelihood signal-to-noise approximation can fail badly. The signal-to -noise ratio of the likelihood ratio itself is also computed for each examp le to demonstrate that it is not a good measure of ideal observer performan ce. (C) 2000 Optical Society of America OCIS codes: 110.3000, 110.2970, 100 .2000, 100.2960, 100.5010.