Figure 2 | Scientific Reports

Figure 2

From: A veracity dissemination consistency-based few-shot fake news detection framework by synergizing adversarial and contrastive self-supervised learning

Figure 2

The model architecture of our proposed DetectYSF, which is a MLM-based prompt learning framework for rumor classification (center). Utilizing the Sentence Representation Contrastive Learning (top-left) and Sample-Level Adversarial Learning (bottom-left) to further enhance the few-shot capability for the prompt-tuning based classification task. The Social Context-based Veracity Dissemination Consistency is then considered via the “veracity feature fusion” strategy during inference stage (right).

Back to article page