False-Positive Rate
Definition
The proportion of authentic media files that a detection tool incorrectly classifies as manipulated or synthetic. Deepfake detectors and ELA tools both have published or empirically derived false-positive rates that must be cited in the report when their outputs are used as evidence.
- Fingerprint FP rate
- Approximately 0.1% (Ulery 2011 study)
- Key metric missing
- PCAST found false-positive rate absent for tool marks
- Primary use
- Basis for rejecting zero-error-rate expert testimony
Common questions
What exactly is a false-positive rate in forensic pattern analysis?+
It is the proportion of pairs from different sources that an examiner incorrectly matches. In other words, it measures how often the examiner says two samples match when they actually come from different sources. The Ulery 2011 fingerprint study measured this at approximately 0.1%.
Why did PCAST say false-positive rate data was missing for tool marks?+
Tool mark examiners did not have empirical studies measuring their false-positive rate, unlike fingerprint examiners. Without this data, there was no way to test whether tool mark identifications could be relied upon to exclude innocent sources.
Does false-positive rate change between different forensic labs or regions?+
Yes. The false-positive rate depends on the statistical method used, the reference population compared against, and the geological diversity of the region. Two labs using the same technique on samples from different areas may get different error rates.
Related terms
- AFTE Theory of Identification
- The Association of Firearm and Tool Mark Examiners' standard (1992, revised 2011) for reporting tool-mark comparison conclusions as Identification, Inconclusive, or Elimination...
- Asperity
- A microscopic surface protrusion on a tool's cutting edge; each asperity produces a groove in the substrate when the tool slides, contributing...
- Black-Box Study
- A study in which qualified examiners are given realistic test cases with known ground truth and asked to make decisions under conditions...
- Black-Box Validation
- A study design where examiners are given unknown questioned-vs-known pairs and asked to match or exclude without feedback, allowing error rates to...
- BSA 2023
- Bharatiya Sakshya Adhiniyam 2023: the Indian evidence statute that replaced the Indian Evidence Act 1872 on 1 July 2024; retains the section...
- C2PA (Coalition for Content Provenance and Authenticity)
- An open technical standard that embeds cryptographically signed provenance assertions into media files at the point of capture or editing. A C2PA...
- Class Characteristics
- Features deliberately or systematically imparted to a group of items by design or manufacture, such as the calibre of a bullet, the...
- CrimPR Part 19
- Criminal Procedure Rules Part 19 (England and Wales): the expert-witness procedural code requiring disclosure of qualifications, methodology, limitations, and whether the method...
- Daubert Standard
- The US federal evidentiary standard (Daubert v. Merrell Dow Pharmaceuticals, 1993) requiring that expert testimony be based on scientifically valid methods with...
- Deepfake Detector
- A machine-learning classifier trained to distinguish authentic recordings from AI-synthesised or face-swapped media. Outputs a probability score rather than a binary verdict....
- Error Level Analysis (ELA)
- A technique that re-compresses a JPEG at a controlled quality setting and maps the pixel-level difference between the re-compressed and original images....
- Evaluative Reporting
- A framework for expressing forensic conclusions in terms of the probability of the evidence given the prosecution hypothesis versus the defence hypothesis...
Explained in these topics
- Expert Testimony and the 2009 NAS Critique of Fingerprint IDThe proportion of non-mate comparisons (different fingers) on which an examiner reaches an identification conclusion; measured at approximately 0.1% in the Ule...
- Statistical Comparison and the Reference Population ProblemThe proportion of cases in which two samples from genuinely different sources are incorrectly classified as sharing a source. Depends on the statistical method...
- Tool Marks: Fundamentals, Striations and ImpressionsIn pattern comparison: the proportion of pairs that are actually from different sources but incorrectly called a match; the key metric PCAST found absent for t...
- Writing a Media Authenticity Examination ReportThe proportion of authentic media files that a detection tool incorrectly classifies as manipulated or synthetic. Deepfake detectors and ELA tools both have pu...