---
title: "Two-stage Training for Learning from Label Proportions"
authors:
  - "Jiabin Liu"
  - "Bo Wang"
  - "Xin Shen"
  - "Zhiquan Qi"
  - "Yingjie Tian"
doi: "10.24963/ijcai.2021/377"
url: "https://ijcai.org/proceedings/2021/377"
pdf: "https://ijcai.org/proceedings/2021/0377.pdf"
venue: "Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence"
year: 2021
pages: "2737-2743"
track: "Main Track"
keywords:
  - "Machine Learning: Weakly Supervised Learning"
  - "Machine Learning: Deep Learning"
  - "Machine Learning: Classification"
---

# Two-stage Training for Learning from Label Proportions

Jiabin Liu, Bo Wang, Xin Shen, Zhiquan Qi, Yingjie Tian

*Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence*, pages 2737-2743. DOI: [10.24963/ijcai.2021/377](https://doi.org/10.24963/ijcai.2021/377)

## Abstract

Learning from label proportions (LLP) aims at learning an instance-level classifier with label proportions in grouped training data. Existing deep learning based LLP methods utilize end-to-end pipelines to obtain the proportional loss with Kullback-Leibler divergence between the bag-level prior and posterior class distributions. However, the unconstrained optimization on this objective can hardly reach a solution in accordance with the given proportions. Besides, concerning the probabilistic classifier, this strategy unavoidably results in high-entropy conditional class distributions at the instance level. These issues further degrade the performance of the instance-level classification. In this paper, we regard these problems as noisy pseudo labeling, and instead impose the strict proportion consistency on the classifier with a constrained optimization as a continuous training stage for existing LLP classifiers. In addition, we introduce the mixup strategy and symmetric cross-entropy to further reduce the label noise. Our framework is model-agnostic, and demonstrates compelling performance improvement in extensive experiments, when incorporated into other deep LLP models as a post-hoc phase.