J Shanghai Jiaotong Univ Sci ›› 2023, Vol. 28 ›› Issue (6): 703-716.doi: 10.1007/s12204-022-2487-5

• • 上一篇    下一篇



  1. (1.上海交通大学 电子信息与电气工程学院,上海200240;2. 上海市数字媒体处理与传输重点实验室,上海200240)
  • 接受日期:2021-07-23 出版日期:2023-11-28 发布日期:2023-12-04

Boosting Unsupervised Domain Adaptation with Soft Pseudo-Label and Curriculum Learning

ZHANG Shengjia(张晟嘉),LIN Tiancheng(林天成),XU Yi*(徐奕)   

  1. (School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China; Shanghai Key Laboratory of Digital Media Processing and Transmission, Shanghai 200240, China)
  • Accepted:2021-07-23 Online:2023-11-28 Published:2023-12-04

摘要: 无监督域适应利用标签完整的源域数据,通过显式的数据分布差异最小化或对抗学习,提高无标签目标域的分类性能。作为一种增强方法,在域适应过程中会涉及类别对齐,即利用模型预测来加强目标特征识别。此方法存在两个问题:在目标域中,错误的类别预测会导致伪标签不准确;在源域中,过拟合会导致分布偏差。因此本文提出了一种与模型无关的两阶段学习框架,利用软伪标签策略大大减少了错误的模型预测,并利用课程学习策略避免了源域的过拟合。理论上,成功降低目标域预期误差上限的综合风险。在第一阶段,我们使用基于分布对齐的无监督域适应方法训练模型,以获得置信度相当高的目标域软语义标签。为了避免源域的过拟合,在第二阶段,我们提出了一种课程学习策略,以自适应性地控制两个域损失之间的权重,从而使训练阶段的重点逐渐从源域分布转移到目标域分布,并提高目标域的预测置信度。在两个常见基准数据集上进行的广泛实验验证了我们提出的框架在提升排名靠前的无监督域适应算法性能方面的普遍有效性,并证明了其一贯的卓越性能。

关键词: 无监督域适应, 伪标签, 软标签, 课程学习

Abstract: By leveraging data from a fully labeled source domain, unsupervised domain adaptation (UDA) improves classification performance on an unlabeled target domain through explicit discrepancy minimization of data distribution or adversarial learning. As an enhancement, category alignment is involved during adaptation to reinforce target feature discrimination by utilizing model prediction. However, there remain unexplored problems about pseudo-label inaccuracy incurred by wrong category predictions on target domain, and distribution deviation caused by overfitting on source domain. In this paper, we propose a model-agnostic two-stage learning framework, which greatly reduces flawed model predictions using soft pseudo-label strategy and avoids overfitting on source domain with a curriculum learning strategy. Theoretically, it successfully decreases the combined risk in the upper bound of expected error on the target domain. In the first stage, we train a model with distribution alignment-based UDA method to obtain soft semantic label on target domain with rather high confidence. To avoid overfitting on source domain, in the second stage, we propose a curriculum learning strategy to adaptively control the weighting between losses from the two domains so that the focus of the training stage is gradually shifted from source distribution to target distribution with prediction confidence boosted on the target domain. Extensive experiments on two well-known benchmark datasets validate the universal effectiveness of our proposed framework on promoting the performance of the top-ranked UDA algorithms and demonstrate its consistent superior performance.