Score Likelihood Ratio

Photo by CSAFE

In forensic settings, Likelihood Ratios and Bayes Factor are used to compute probative value of evidence when assessing two opposing propositions. However, these methods require formulating a probability model that may be too complex or infeasible to estimate. The use of score-based likelihood ratios or SLR has been proposed as an alternative for this problem. Our work is center in two issues:

  • Evaluating the use of machine learning based SLR for forensic evidence.
  • Assessing the effect of dependence structured on SLR data and propose solutions.

I have worked on both project with Dr.Danica Ommen.

1. Evaluation of Score-based Likelihood Ratios in a Forensic Setting.

a.Evaluation of Score-based Likelihood Ratios in a Forensic Setting [Updated version submitted for publication].

b. Previous versions

  • Poster presentations:

2. Ensemble of SLR systems for forensic evidence.

  • Publications:
  • Poster presentations:
  • Talks:
Federico Veneri Guarch
Federico Veneri Guarch
Research Assistant