一作ACL 2026Poster· Natural Language Processing
Adversarial Metric Learning for Fine-Grained Emotion Classification
Junfan Chen · Sizhe Wu · Richong Zhang · Chunming Hu
摘要
Fine-grained emotion classification (FEC) requires distinguishing subtly different emotions, where the dominant errors come from closely confusable categories. Recent progress relies on contrastive learning with hard-pair mining, implicitly assuming that a fixed similarity metric is sufficient to...