Multimodal Deep Learning Framework for Forensic Emotion and Behavior Signal Analysis
摘要
This paper advocates for a transformative shift in forensic science by leveraging artificial intelligence (AI)-enabled behavioral biometrics as a novel forensic detection methodology. Traditional forensic methods, though foundational, often face significant limitations in complex investigative scenarios involving incomplete or compromised physical evidence. To address this gap, we propose integrating three distinct behavioral biometric techniques—micro-expression recognition, gait analysis, and digital behavioral pattern profiling—into a unified forensic framework powered by advanced explainable deep learning algorithms. This paper highlights the unique advantages of behavioral biometrics, emphasizing their robustness against deception and concealment attempts, thus enhancing forensic accuracy and reliability. Additionally, it identifies critical challenges, including ethical considerations and legal admissibility, calling for multidisciplinary collaboration. The potential benefits and transformative impact presented by this AI-driven approach underscore the urgency for further exploration and practical implementation in modern forensic science.