Approximations to ideal-observer performance on signal-detection tasks

Research output: Contribution to journalArticle

26 Citations (Scopus)

Abstract

The ideal-observer performance, as measured by the area under the receiver's operating characteristic curve, is computed for six examples of signal-detection tasks. Exact values for this quantity, as well as approximations based on the signal-to-noise ratio of the log likelihood and the likelihood-generating function, are found. The noise models considered are normal, exponential, 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 example to demonstrate that it is not a good measure of ideal-observer performance.

Original languageEnglish (US)
Pages (from-to)1783-1793
Number of pages11
JournalApplied Optics
Volume39
Issue number11
StatePublished - Apr 10 2000

Fingerprint

signal detection
Signal detection
Signal to noise ratio
signal to noise ratios
approximation
likelihood ratio
receivers
curves

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics

Cite this

Approximations to ideal-observer performance on signal-detection tasks. / Clarkson, Eric W; Barrett, Harrison H.

In: Applied Optics, Vol. 39, No. 11, 10.04.2000, p. 1783-1793.

Research output: Contribution to journalArticle

@article{bdbdf0b9b180480a8f9232aa416f19c7,
title = "Approximations to ideal-observer performance on signal-detection tasks",
abstract = "The ideal-observer performance, as measured by the area under the receiver's operating characteristic curve, is computed for six examples of signal-detection tasks. Exact values for this quantity, as well as approximations based on the signal-to-noise ratio of the log likelihood and the likelihood-generating function, are found. The noise models considered are normal, exponential, 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 example to demonstrate that it is not a good measure of ideal-observer performance.",
author = "Clarkson, {Eric W} and Barrett, {Harrison H}",
year = "2000",
month = "4",
day = "10",
language = "English (US)",
volume = "39",
pages = "1783--1793",
journal = "Applied Optics",
issn = "1559-128X",
publisher = "The Optical Society",
number = "11",

}

TY - JOUR

T1 - Approximations to ideal-observer performance on signal-detection tasks

AU - Clarkson, Eric W

AU - Barrett, Harrison H

PY - 2000/4/10

Y1 - 2000/4/10

N2 - The ideal-observer performance, as measured by the area under the receiver's operating characteristic curve, is computed for six examples of signal-detection tasks. Exact values for this quantity, as well as approximations based on the signal-to-noise ratio of the log likelihood and the likelihood-generating function, are found. The noise models considered are normal, exponential, 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 example to demonstrate that it is not a good measure of ideal-observer performance.

AB - The ideal-observer performance, as measured by the area under the receiver's operating characteristic curve, is computed for six examples of signal-detection tasks. Exact values for this quantity, as well as approximations based on the signal-to-noise ratio of the log likelihood and the likelihood-generating function, are found. The noise models considered are normal, exponential, 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 example to demonstrate that it is not a good measure of ideal-observer performance.

UR - http://www.scopus.com/inward/record.url?scp=0003844860&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0003844860&partnerID=8YFLogxK

M3 - Article

C2 - 18345075

AN - SCOPUS:0003844860

VL - 39

SP - 1783

EP - 1793

JO - Applied Optics

JF - Applied Optics

SN - 1559-128X

IS - 11

ER -