A New Source Detection Algorithm Using the False-Discovery Rate | Zendy

Andrew Hopkins | Zendy; C. Miller | Zendy; Andrew J. Connolly | Zendy; Christopher R. Genovese | Zendy; R. C. Nichol | Zendy; Larry Wasserman | Zendy

AI Assistant Blog Pricing

Home ZAIA Blog

Open Access

A New Source Detection Algorithm Using the False-Discovery Rate

Author(s) -

Andrew Hopkins,

C. Miller,

Andrew J. Connolly,

Christopher R. Genovese,

R. C. Nichol,

Larry Wasserman

Publication year - 2002

Publication title -

the astronomical journal

Language(s) - English

Resource type - Journals

SCImago Journal Rank - 2.61

H-Index - 271

eISSN - 1538-3881

pISSN - 0004-6256

DOI - 10.1086/338316

Subject(s) - false discovery rate , false positive paradox , fraction (chemistry) , pixel , computer science , algorithm , a priori and a posteriori , constraint (computer aided design) , multiple comparisons problem , false positives and false negatives , false positive rate , pattern recognition (psychology) , artificial intelligence , data mining , statistics , mathematics , biochemistry , chemistry , philosophy , geometry , organic chemistry , epistemology , gene

The False Discovery Rate (FDR) method has recently been described by Milleret al (2001), along with several examples of astrophysical applications. FDR isa new statistical procedure due to Benjamini and Hochberg (1995) forcontrolling the fraction of false positives when performing multiple hypothesistesting. The importance of this method to source detection algorithms isimmediately clear. To explore the possibilities offered we have developed a newtask for performing source detection in radio-telescope images, Sfind 2.0,which implements FDR. We compare Sfind 2.0 with two other source detection andmeasurement tasks, Imsad and SExtractor, and comment on several issues arisingfrom the nature of the correlation between nearby pixels and the necessaryassumption of the null hypothesis. The strong suggestion is made thatimplementing FDR as a threshold defining method in other existingsource-detection tasks is easy and worthwhile. We show that the constraint onthe fraction of false detections as specified by FDR holds true even for highlycorrelated and realistic images. For the detection of true sources, which arecomplex combinations of source-pixels, this constraint appears to be somewhatless strict. It is still reliable enough, however, for a priori estimates ofthe fraction of false source detections to be robust and realistic.Comment: 17 pages, 7 figures, accepted for publication by A

The content you want is available to Zendy users.

Already have an account? Click here to sign in.

Having issues? You can contact us here

Accelerating Research