Representation Learning and Data Augmentation in Forensic Comparison

• NCJ Number: 310339
• Author(s): Jun Chu
• Date Published: 2018
• Length: 105 pages
• Grant Number(s): 2017-MU-CX-0010
• Agencies: NIJ-Sponsored
• Publication Type: Thesis/Dissertation

Annotation

In this study, the authors introduce a discriminative method based on representation and data augmentation to address the forensic comparison problem.

Abstract

To overcome the forensic comparison problem, the authors of this study propose a discriminative method based on automatically learned image representations and data augmentation. The forensic comparison problem is about determining whether observed evidence arises from a known source by comparing the evidence with known samples side by side. The opinion of a forensic comparison system is characterized by the likelihood ratio (LR), which has been previously calculated using generative methods. However, due to intractability of generative models in high-dimensional spaces, previous methods collapse inputs to a distance scalar based on human-engineered feature vectors. The drawbacks are the huge loss of information and the imperfection of human-engineered features. The authors use the handwritten images as an example of the forensic comparison problem to demonstrate that by combining unsupervised representation learning, supervised representation learning and data augmentation, one can implement a hybrid system that outperforms previous methods based on human-engineered features. The authors generalize the one-to-one forensic comparison problem to many-to-many comparisons and writer-independent one-to-many comparisons by comparing distributions formed by learned image representations in the latent space. This is done by designing and extracting parametric statistical features from those distributions. The statistical features on sample distributions are expected to capture both within-writer variability and cross-writer variability. As a baseline, the authors compare the performance of human-engineered image extractors such as the gradient, structural and concavity (GSC) micro features with their automatically learned counterpart. Results show that the auto-learned features perform better. The authors also compare distribution comparison using distance-based non-parametric methods (K-S test) with parametric statistical features extracted from image feature vectors. Experiments show that the latter trains much faster than the former and yields better verification accuracies.