研究人员开发了改进多标签分类任务的新方法，该任务涉及为单个实例预测多个标签。一种名为RAPT的方法，作为一种模型无关的包装器，通过检索相似的过往案例来调整标签选择阈值，其表现优于静态阈值设置和少样本LLM。另一个框架PIAA增强了斑块级推理，并使用自适应聚合进行多标签图像识别，在无需重新训练的情况下取得了显著的进步。此外，还提出了一个用于优化多标签学习中广义度量指标的理论框架，提供了具有可证明保证的原则性算法，并在大型数据集上展示了可扩展性。 AI
影响 这些进展为复杂的分类问题提供了更强大、更高效的解决方案，有可能在文档理解和图像识别等领域提高性能。
排序理由 该集群包含多篇学术论文，详细介绍了机器学习任务（特别是多标签分类和识别）的新算法和框架。
在 arXiv cs.IR (Information Retrieval) 阅读 →
AI 生成摘要 · Google Gemini · 来自 5 个来源。 我们如何撰写摘要 →
Industrial multi-label document understanding pipelines score candidate labels and threshold or rank them to form a label set per document. This early selection step directly affects the accuracy of downstream information extraction from the document, as well as the associated ve…
arXiv:2605.28767v1 Announce Type: cross Abstract: Many real-world classification tasks require predicting multiple labels per instance, necessitating the optimization of complex evaluation metrics such as the $F$-measure and Jaccard index. While the Empirical Utility Maximization…
Many real-world classification tasks require predicting multiple labels per instance, necessitating the optimization of complex evaluation metrics such as the $F$-measure and Jaccard index. While the Empirical Utility Maximization (EUM) framework is natural for these population-l…
arXiv:2605.25821v1 Announce Type: new Abstract: Vision-Language Models such as CLIP exhibit strong zero-shot recognition capability by aligning images with textual concepts, yet they often underperform on multi-label recognition where multiple objects co-exist. A key bottleneck i…
Vision-Language Models such as CLIP exhibit strong zero-shot recognition capability by aligning images with textual concepts, yet they often underperform on multi-label recognition where multiple objects co-exist. A key bottleneck is that the [CLS] token, as a single global visua…