On determining the radar threshold for non-Gaussian processes from experimental data

Rejection Threshold Estimation for an Unknown Language Model in an OCR Task

“…In sensor systems, where large amounts of data are usually available, the target detection is seriously affected by false positives, and a special effort has been made to improve their behavior. Thus, Ozturk et al [14] used the generalized Pareto distribution to approximate the extreme tail of the distributions of radar measures, and propose the ordered sample least squares method for estimating the parameters of the distributions. Recently, Broadwater and Chellappa [4] proposed an algorithm using extreme value theory through the use of the generalized Pareto distribution, too, and a Kolmogorov-Smirnov statistical test, and propose a way to adaptively maintain low false positive rates and to overcome differences between the model assumptions and the real data.…”

Aplicaciones de los autómatas transductores finitos con pesos (WFST) en la corrección simbólica en interfaces persona-máquina

“…En el campo de la detección de la señal, se han realizado extensos trabajos durante décadas entre los que cabe citar, Ozturk et al [1996]. En este trabajo se utiliza la distribución generalizada de Pareto para aproximar las colas de la distribución de las medidas de un radar, y se propone el método de mínimos cuadrados de muestras ordenadas (OSLS) para estimar el umbral.…”