MDPE: A Multimodal Deception Dataset with Personality and Emotional Characteristics
Cong Cai, Shan Liang, Xuefei Liu, Kang Zhu, Zhengqi Wen, Jianhua Tao, Heng Xie, Jizhou Cui, Yiming Ma, Zhenhua Cheng, Hanzhe Xu, Ruibo Fu, Bin Liu, Yongwei Li
TL;DR
This paper introduces MDPE, a comprehensive multimodal deception dataset with personality and emotional data, enabling improved deception detection and related affective computing research.
Contribution
The paper presents MDPE, a novel dataset that includes multimodal deception, personality, and emotional data, addressing the lack of datasets for evaluating individual differences in deception detection.
Findings
MDPE contains over 104 hours of videos from 193 subjects.
Experiments demonstrate MDPE's utility for deception, personality, and emotion recognition.
The dataset facilitates exploring relationships between deception and individual differences.
Abstract
Deception detection has garnered increasing attention in recent years due to the significant growth of digital media and heightened ethical and security concerns. It has been extensively studied using multimodal methods, including video, audio, and text. In addition, individual differences in deception production and detection are believed to play a crucial role.Although some studies have utilized individual information such as personality traits to enhance the performance of deception detection, current systems remain limited, partly due to a lack of sufficient datasets for evaluating performance. To address this issue, we introduce a multimodal deception dataset MDPE. Besides deception features, this dataset also includes individual differences information in personality and emotional expression characteristics. It can explore the impact of individual differences on deception…
Peer Reviews
Decision·Submitted to ICLR 2025
The new dataset can be valuable to the community if made publicly available. The authors present extensive evaluations of existing methods/features on the dataset.
The concept of emotion and emotion expression used in the paper is fuzzy. E.g. what is a "true emotional expression" (line 074) supposed to refer to? One possibility that I might suspect is that the authors refer to whether the emotional expression is aligned with an internal state. This is a highly complex topic as emotions are not displayed directly on e.g. a person's face but are subject to regulation processes and social display rules (e.g. see Schneeberger et al., 2023; Müller et al., 2024)
The authors conducted the data collection with great effort and provided comprehensive coverage regarding the data collection and label design.
1. The 'Introduction' section is somewhat verbose. While the authors did an excellent job of providing detailed background information on deception with examples from various modalities, I believe that some of this content, particularly the details in the second and third paragraphs, would be better suited for the 'Related Work' section. 2. The authors mention "effective incentives" without clarifying what these entail (whether monetary or non-monetary) or how they were distributed. Additionall
The topic of deception detection is a challenging and important one.
My main concerns with this paper are about the novelty, the content and the presentation style. Regarding the content, the contribution of this paper seems quite limited for a top conference on machine learning (alternative and potentially more adequate conferences could be ICMI. CSCW, ACM-MM, etc.). Regarding the presentation, there are several typos and some sections of methods (Section 4.2 for example) are very badly organized and written in a not adequate manner.
Code & Models
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsCybercrime and Law Enforcement Studies · Deception detection and forensic psychology · Crime Patterns and Interventions
MethodsSoftmax · Attention Is All You Need