Deep Learning for Forensic Identification of Source

2025-03-26

AI Summary

This work addresses the core forensic question: “Do two cartridge cases originate from the same unknown firearm?”—for the first time directly applying contrastive learning to cartridge case source attribution. Leveraging the NBIDE and E3 cartridge case image datasets, we design an end-to-end contrastive neural network that learns interpretable pairwise similarity scores, enabling likelihood ratio interpretation under the “common but unknown source” paradigm. Experiments on the E3 dataset yield a ROC AUC of 0.892, significantly surpassing state-of-the-art CMC-based methods (0.867). Ablation studies confirm strong robustness to variations in network width and depth. To our knowledge, this is the first deep contrastive learning framework for forensic ballistic analysis that simultaneously achieves high accuracy, interpretability, and principled evidential reasoning—providing a novel, reliable solution for firearms identification in casework.

Technology Category

Machine Learning: Calibration & Uncertainty QuantificationCognitive Modeling & Cognitive Systems: Conceptual Inference and ReasoningComputer Vision: Adversarial Attacks & Robustness

Application Category

Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingUser Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systemsWeb Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

We used contrastive neural networks to learn useful similarity scores between the 144 cartridge casings in the NBIDE dataset, under the common-but-unknown source paradigm. The common-but-unknown source problem is a problem archetype in forensics where the question is whether two objects share a common source (e.g. were two cartridge casings fired from the same firearm). Similarity scores are often used to interpret evidence under this paradigm. We directly compared our results to a state-of-the-art algorithm, Congruent Matching Cells (CMC). When trained on the E3 dataset of 2967 cartridge casings, contrastive learning achieved an ROC AUC of 0.892. The CMC algorithm achieved 0.867. We also conducted an ablation study where we varied the neural network architecture; specifically, the network's width or depth. The ablation study showed that contrastive network performance results are somewhat robust to the network architecture. This work was in part motivated by the use of similarity scores attained via contrastive learning for standard evidence interpretation methods such as score-based likelihood ratios.

Problem

• Develop contrastive neural networks for forensic source identification
• Compare performance with state-of-the-art Congruent Matching Cells algorithm
• Assess robustness of network architecture in similarity scoring

Innovation

• Contrastive neural networks for similarity scores
• Comparison with Congruent Matching Cells algorithm
• Ablation study on network architecture robustness