timezone icon TimeZone (UTC)
A General Framework for the Estimation of Likelihood Ratios
50 Minutes
This talk provides a very general framework for the estimation of likelihood ratios using similarity or dissimilarity scores resulting from a comparison of two patterns that overcome this deficiency for suitably chosen similarity measures. In such instances, the likelihood ratio estimate defined is in fact a good approximation for the classical likelihood ratio. Moreover, the quality of the approximation improves with increasing empirical information. This approach can be applied even when the data are numbers, vectors or attributes such as color. This presentation illustrates the approach using examples involving glass fragments, fingerprint comparisons, bullet casing marks and DNA profiles.
This webinar was recorded in its entirety at the time of the live event in order to capture the one-on-one interaction with the presenter.


Steve Lund
Statistical Engineering Division, Information Technology Laboratory, National Institute of Standards and Technology.

This event is funded through a Cooperative Agreement (2011-DN-BX-K564) from the National Institute

of Justice (NIJ), Office of Justice Programs (OJP), and U.S. Department of Justice (USDOJ). The views,

policies, and opinions expressed are those of the authors and contributors and do not necessarily reflect

those of the NIJ, OJP, or USDOJ.