Explainable artificial intelligence in forensics: Realistic explanations for number of contributor predictions of DNA profiles

Veldhuis, Marthe S.; Ariëns, Simone; Ypma, Rolf J. F.; Abeel, Thomas; Benschop, Corina C.G.

doi:10.1016/j.fsigen.2021.102632

Cited by 13 publications

(3 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to the BlackBox nature of DL algorithms as well as due to growing complexities, the need for explainability is increasing rapidly, especially in image processing [ 42 , 43 , 44 ], criminal investigation [ 45 , 46 ], forensic [ 47 , 48 , 49 ], etc. Professionals from these sectors may find it easier to comprehend the DL model’s findings and apply them to swiftly and precisely assess whether a face is real or artificial.…”

Section: Methodsmentioning

confidence: 99%

Auguring Fake Face Images Using Dual Input Convolution Neural Network

et al. 2022

View full text Add to dashboard Cite

Deepfake technology uses auto-encoders and generative adversarial networks to replace or artificially construct fine-tuned faces, emotions, and sounds. Although there have been significant advancements in the identification of particular fake images, a reliable counterfeit face detector is still lacking, making it difficult to identify fake photos in situations with further compression, blurring, scaling, etc. Deep learning models resolve the research gap to correctly recognize phony images, whose objectionable content might encourage fraudulent activity and cause major problems. To reduce the gap and enlarge the fields of view of the network, we propose a dual input convolutional neural network (DICNN) model with ten-fold cross validation with an average training accuracy of 99.36 ±0.62, a test accuracy of 99.08 ± 0.64, and a validation accuracy of 99.30 ± 0.94. Additionally, we used ’SHapley Additive exPlanations (SHAP) ’ as explainable AI (XAI) Shapely values to explain the results and interoperability visually by imposing the model into SHAP. The proposed model holds significant importance for being accepted by forensics and security experts because of its distinctive features and considerably higher accuracy than state-of-the-art methods.

show abstract

Section: Methodsmentioning

confidence: 99%

Auguring Fake Face Images Using Dual Input Convolution Neural Network

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Allele sharing among contributors to a mixture and masking of alleles due to STR stutter artifacts can lead to inaccurate NoC estimates based on simply counting the number of alleles at a locus. Different approaches and software programs have been used for NoC estimation [ [269] , [270] , [271] , [272] , [273] , [274] , [275] ]. Total allele count (TAC) distribution via TAC curves showed an improvement in manually estimating the number of contributors with complex mixtures [ 276 ].…”

Section: Advancements In Current Practicesmentioning

confidence: 99%

Recent advances in forensic biology and forensic DNA typing: INTERPOL review 2019–2022

Butler

2023

Forensic Science International: Synergy

View full text Add to dashboard Cite

“…This exercise is challenging, and the question remains whether a jury can draw a reasonable adverse inference. For these purposes, machine learning could be an optimal tool to evaluate the number of contributors in mixed profiles [97], as well as in the evaluation of complex Bayesian networks [91]. As regards these considerations, it should be taken into account that to date, the court is not always prepared to receive and interpret this kind of report to give the right "weight of evidence".…”

mentioning

confidence: 99%

Indirect DNA Transfer and Forensic Implications: A Literature Review

Sessa,

Pomara,

Esposito

et al. 2023

Genes

View full text Add to dashboard Cite

Progress in DNA profiling techniques has made it possible to detect even the minimum amount of DNA at a crime scene (i.e., a complete DNA profile can be produced using as little as 100 pg of DNA, equivalent to only 15–20 human cells), leading to new defense strategies. While the evidence of a DNA trace is seldom challenged in court by a defendant’s legal team, concerns are often raised about how the DNA was transferred to the location of the crime. This review aims to provide an up-to-date overview of the experimental work carried out focusing on indirect DNA transfer, analyzing each selected paper, the experimental method, the sampling technique, the extraction protocol, and the main results. Scopus and Web of Science databases were used as the search engines, including 49 papers. Based on the results of this review, one of the factors that influence secondary transfer is the amount of DNA shed by different individuals. Another factor is the type and duration of contact between individuals or objects (generally, more intimate or prolonged contact results in more DNA transfer). A third factor is the nature and quality of the DNA source. However, there are exceptions and variations depending on individual characteristics and environmental conditions. Considering that secondary transfer depends on multiple factors that interact with each other in unpredictable ways, it should be considered a complex and dynamic phenomenon that can affect forensic investigation in various ways, for example, placing a subject at a crime scene who has never been there. Correct methods and protocols are required to detect and prevent secondary transfer from compromising forensic evidence, as well as the correct interpretation through Bayesian networks. In this context, the definition of well-designed experimental studies combined with the use of new forensic techniques could improve our knowledge in this challenging field, reinforcing the value of DNA evidence in criminal trials.

show abstract

Explainable artificial intelligence in forensics: Realistic explanations for number of contributor predictions of DNA profiles

Cited by 13 publications

References 50 publications

Auguring Fake Face Images Using Dual Input Convolution Neural Network

Auguring Fake Face Images Using Dual Input Convolution Neural Network

Recent advances in forensic biology and forensic DNA typing: INTERPOL review 2019–2022

Indirect DNA Transfer and Forensic Implications: A Literature Review

Contact Info

Product

Resources

About