Skip to content

PROBABILITY AND STATISTICS FOR PATTERN AND
DIGITAL EVIDENCE

Probability and Statistics for Pattern and Digital Evidence

All forms of evidence have some degree of uncertainty. This uncertainty is often misunderstood in criminal cases. Faulty evidence, or invalid inferences derived from the evidence, leads to costly mistakes and loss of confidence in the criminal justice system.

Current challenges facing the forensic and legal communities include:

  • Insufficient scientifically valid methods to analyze/interpret pattern or digital evidence
  • Lack of representative databases to develop new methods or to test existing ones
  • Few well-designed studies to estimate error rates, including those of technologies used in court daily

Prior to 2015, pattern and digital disciplines saw limited progress in terms of statistical thinking. That is why CSAFE promotes a rigorous research agenda focused on:

  • Development and implementation of new probabilistic methods and statistical tools
  • Rigorous evaluations of existing methods and tools
  • Collection and dissemination of publicly-available data sets to the broader forensic and research communities

Pattern Evidence

Evidence analysis of pattern comparison disciplines, such as firearms examination, latent print analysis, and footwear impressions, involves 2D images or 3D impressions. When digitized, they result in massive image files, making it difficult to extract meaningful features.

Comparing items through a visual inspection then expressing a subjective opinion about the degree of similarity between them is still an industry best practice. While proficient, careful examiners will often reach the correct conclusion, it is vital to develop objective comparison approaches with estimable error rates. This is the focus of CSAFE research in pattern evidence analysis.

Digital Evidence

Digital evidence is growing in importance as digital devices become ubiquitous in modern society. This type of evidence poses unique challenges, from recovering illicit data or software hidden on phones and tablets to examining digital traces of user activity (e.g., messages/texts, login events) for signs of criminal action. CSAFE research addresses the increased demand for digital forensics tools and methods, all built on strong probabilistic and statistics foundations. This is the focus of CSAFE research in digital evidence analysis.