DecepGPT: 基于多文化数据集和鲁棒多模态学习的模式驱动欺骗检测
DecepGPT: Schema-Driven Deception Detection with Multicultural Datasets and Robust Multimodal Learning
AI总结：
本文提出DecepGPT，通过构建包含结构化线索描述和推理链的推理数据集，释放多文化数据集T4-Deception，并提出SICS和DMC模块，实现多模态欺骗检测的鲁棒学习，实验表明其在领域内和跨领域场景中均取得最佳性能。
摘要：
多模态欺骗检测旨在通过分析音频视觉线索来识别欺骗行为，用于刑侦和安全领域。在高风险环境中，调查人员需要可验证的证据将音频视觉线索与最终决策联系起来，并且需要在不同领域和文化背景下可靠地泛化。然而，现有基准仅提供二元标签而无中间推理线索。数据集也较小，场景覆盖有限，导致捷径学习。我们通过三个贡献解决这些问题：首先，我们通过增强现有基准并添加结构化线索级描述和推理链来构建推理数据集，使模型输出可审计报告。其次，我们发布T4-Deception，一个基于统一的『To Tell The Truth』电视格式在四个国家实施的多文化数据集。该数据集包含1695个样本，是目前最大的非实验室欺骗检测数据集。第三，我们提出两个模块，以在小数据条件下实现鲁棒学习。Stabilized Individuality-Commonality Synergy (SICS) 通过结合可学习的全局先验与样本自适应残差，优化多模态表示，随后通过极性感知调整双向校准表示。Distilled Modality Consistency (DMC) 通过知识蒸馏将模态特定预测与融合的多模态预测对齐，以防止单模态捷径学习。在三个已建立的基准和我们新的数据集上的实验表明，我们的方法在领域内和跨领域场景中均取得最佳性能，同时在不同文化背景下表现出优越的迁移能力。数据集和代码将被发布。
英文摘要：
Multimodal deception detection aims to identify deceptive behavior by analyzing audiovisual cues for forensics and security. In these high-stakes settings, investigators need verifiable evidence connecting audiovisual cues to final decisions, along with reliable generalization across domains and cultural contexts. However, existing benchmarks provide only binary labels without intermediate reasoning cues. Datasets are also small with limited scenario coverage, leading to shortcut learning. We address these issues through three contributions. First, we construct reasoning datasets by augmenting existing benchmarks with structured cue-level descriptions and reasoning chains, enabling models to output auditable reports. Second, we release T4-Deception, a multicultural dataset based on the unified ``To Tell the Truth'' television format implemented across four countries. With 1695 samples, it is the largest non-laboratory deception detection dataset. Third, we propose two modules for robust learning under small-data conditions. Stabilized Individuality-Commonality Synergy (SICS) refines multimodal representations by combining learnable global priors with sample-adaptive residuals and applying polarity-aware recalibration. Distilled Modality Consistency (DMC) aligns modality-specific predictions with the fused multimodal predictions via knowledge distillation to prevent unimodal shortcut learning. Experiments on three established benchmarks and our novel dataset demonstrate that our method achieves state-of-the-art performance in both in-domain and cross-domain scenarios, while exhibiting superior transferability across diverse cultural contexts. The datasets and code are available at this link.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
