Extending the scope of probability theory: assessing forensic evidence with belief functions

Duration: 32 mins 3 secs
Share this media item:
Embed this media item:


About this item
Image inherited from collection
Description: Meester, R (Vrije Universiteit Amsterdam)
Thursday 1st September 2016 - 14:30 to 15:00
 
Created: 2016-09-06 12:21
Collection: Probability and Statistics in Forensic Science
Publisher: Isaac Newton Institute
Copyright: Meester, R
Language: eng (English)
Distribution: World     (downloadable)
Explicit content: No
Aspect Ratio: 16:9
Screencast: No
Bumper: UCS Default
Trailer: UCS Default
 
Abstract: The nature of the questions arising in court that can be addressed via probability and statistics also depends on the mathematical machinery we have at our disposal. I do not believe that the classical concept of a probability measures is suitable to capture the kind of uncertainty usually encountered in court cases. For instance, disbelieve in one proposition should not necessarily entail believe in its complement. In this lecture I will revive the old concept of a belief function both philosophically, mathematically, and practically. I will argue that belief functions can be interpreted behaviorally which makes them very suitable mathematical objects to deal with epistemic probability (the kind of probability envountered in court). Then I will explain that it is possible to develop a (new) calculus of belief functions generalizing the classical calculus of probabilities in a very natural way, and finally I will present some applications of belief functions to the classical island problem and to the practical problem of parental identification. The use of belief functions extends the class of question that can be addressed with probabilistic methods in court in a meaningful and practical way, and also helps solving problems with choosing appropriate priors. Indeed, belief functions allow zero prior belief in a meaningful way.
Available Formats
Format Quality Bitrate Size
MPEG-4 Video 640x360    1.94 Mbits/sec 466.59 MB View Download
WebM 640x360    565.21 kbits/sec 132.68 MB View Download
iPod Video 480x270    521.92 kbits/sec 122.52 MB View Download
MP3 44100 Hz 249.83 kbits/sec 58.68 MB Listen Download
Auto * (Allows browser to choose a format it supports)