提出新损失函数，让模型更好识别无标签数据中的隐含正例。
Any-Class Presence Likelihood for Robust Multi-Label Classification with Abundant Negative Data
- 用加权几何平均构建任一类存在的概率，替代传统多标签损失。
- 在多个数据集上提升F1、F2和平均精度，最高增益达6.01个百分点。
- 无需额外参数或计算量，适合工业缺陷、医疗诊断等负样本多场景。
多标签分类（MLC）将实例分配给一个或多个非互斥类别。当数据集中存在大量未标注类别的实例（即负样本）时，会严重干扰学习过程，影响正例的准确识别与分类。然而在工业缺陷检测、农业病害识别和医疗诊断等应用中，负样本极为常见。若为负样本单独设置类别，会增加学习目标复杂度并引入冗余。为此，本文重新设计标准的MLC损失函数，通过预测类别概率的归一化加权几何平均，构造“任一类存在”的似然值，并引入调节参数控制负类概率对正例中该似然的贡献。该似然值增强了网络对隐含正例的感知能力，提升了正例内部的标签分类性能。在包含大量负样本的大规模数据集SewerML、修改版COCO和ChestX-ray14上，使用多种网络和基础损失函数的实验表明，所提损失函数始终优于标准损失，F1提升最高达6.01个百分点，F2提升8.06，平均精度提升3.11，且不增加参数或计算开销。代码已公开于：https://github.com/ML-for-Sensor-Data-Western/gmean-mlc
原文摘要 · Abstract (English)
Multi-label Classification (MLC) assigns an instance to one or more non-exclusive classes. A challenge arises when the dataset contains a large proportion of instances with no assigned class, referred to as negative data, which can overwhelm the learning process and hinder the accurate identification and classification of positive instances. Nevertheless, it is common in MLC applications such as industrial defect detection, agricultural disease identification, and healthcare diagnosis to encounter large amounts of negative data. Assigning a separate negative class to these instances further complicates the learning objective and introduces unnecessary redundancies. To address this challenge, we redesign standard MLC loss functions by deriving a likelihood of any class being present, formulated by a normalized weighted geometric mean of the predicted class probabilities. We introduce a regularization parameter that controls the relative contribution of the absent class probabilities to the any-class presence likelihood in positive instances. The any-class presence likelihood complements the multi-label learning by encouraging the network to become more aware of implicit positive instances and improve the label classification within those positive instances. Experiments on large-scale datasets with negative data: SewerML, modified COCO, and ChestX-ray14, across various networks and base loss functions show that our loss functions consistently improve MLC performance of their standard loss counterparts, achieving gains of up to 6.01 percentage points in F1, 8.06 in F2, and 3.11 in mean average precision, all without additional parameters or computational complexity. Code available at: https://github.com/ML-for-Sensor-Data-Western/gmean-mlc
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。