Fingerprint Evidence

By: Prof. Christophe Champod, Université de Lausanne
WWW: Anders L Madsen

17 February 2019

Introduction

This Bayesian network is based on the statistical data of general patterns of fingerprints on the hands of males and females respectively. It is based on the National Crime Information Centre (NCIC) encoding of respectively the ten-print cards of 17,951,192 males and 4,313,521 females. These datasets have been compiled by the FBI and National Institute of Justice (NIST) and obtained by Andres J. Washington. They can be seen at http://www.dermatoglyphics.com/mfre/ and http://www.dermatoglyphics.com/femfre/.

The general patterns and associated measures (ridge counts for loops and ridge tracing for whorls) are based on the Galton-Henry classification used at the time by the FBI (for more detailed refer to United States Department of Justice, Federal Bureau of Investigation. The Science of Fingerprints. Washington DC: U.S. Government Printing Office, 1984, http://www.gutenberg.org/ebooks/19022.

This network allows computing probabilities associated with various situations of forensic interest. For example, an examiner may want to know the statistical distribution of the general patterns for a given finger number (1 to 5 for the right hand and 6 to 10 for the left hand, from thumb to little finger) and sex. Or alternatively if he/she observes a fingerprint with given attributes (in terms of general pattern and associated measures), he may want to predict the finger number or the sex.

In the first case, it will allow the examiner to appreciate the relative rarity of general patterns on fingerprints. In the second case, it will help making a selection of the best fingers to search for based on the pattern left by unknown source.

A web interface supporting the first decision of this network may look like this:

Select Input

Please select the appropriate state for each of the input nodes listed above.

Issues

Observations

Observations



Case 1: You observe a print that is an inside traced double loop (both these attributes are selected), then looking at the distribution associated the finger number, you know that the most likely finger at its source is a left thumb (finger 6 with 60.70%) followed by the right index (finger 2 with 11.73%). There is a higher probability for male versus female (56.57% over 43.43%).

Load the example:

Case 2 : You want to know the distribution of the general patterns (including ridge counts and ridge tracing) on all male right fingers. The states (Male and Right Hand) are selected (in red) and the probabilities associated with the observations are automatically updated. The most likely general pattern is a right loop (58%), followed by a plain whorl (24.39%).

Load the example:


Acknowledgements

We are grateful to Andres J. Washington (Fingerprint Geometric Analysis, New York) for making public the NCIC datasets used to inform the present Bayesian network.

Contact Information

For further details on the model: Prof. Christophe Champod, Université de Lausanne

For further details on the use of Bayesian networks and web deployment of models contact: Anders L Madsen (alm(at)hugin(dot)com)

References

[Champod et al, 2016] Champod C, Lennard CJ, Margot PA, Stoilovic M. Fingerprints and Other Ridge Skin Impressions. 2nd edition ed. Boca Raton: CRC Press, 2016, chapter 2, pp. 56-65.

[Madsen et al, 2013] Madsen, A. L., Karlsen, M., Barker, G. C., Garcia, A. B., Hoorfar, J., Jensen, F (2013). A Software Package for Web Deployment of Probabilistic Graphical Models. In Proceedings of the Twelfth Scandinavian Conference on Artificial Intelligence (SCAI), pages 175-184.

Useful references for those interested in BBN include:

[Kjærulff and Madsen, 2013] Kjærulff, U. B. and Madsen, A. L. (2013) Bayesian Networks and Influence Diagrams: A Guide to Construction and Analysis. Springer, Second Edition.

Disclaimer
HUGIN EXPERT A/S takes no responsibility whatsoever for examples and information in examples published on this web site. ALL EXAMPLES ARE FOR DEMONSTRATION PURPOSES ONLY.