arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2209.06993cs.CV
收藏

向未来学习:一种新的语义分割自训练框架

Learning from Future: A Novel Self-Training Framework for Semantic Segmentation

Beihang University(北京航空航天大学);The Chinese University of Hong Kong(香港中文大学);Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所);SenseTime Research(商汤科技研究院);Shanghai Jiao Tong University(上海交通大学)

Ye Du, Yujun Shen, Haochen Wang, Jingjing Fei, Wei Li, Liwei Wu, Rui Zhao, Zehua Fu, Qingjie Liu

摘要:

自我训练在半监督学习中显示出巨大的潜力。它的核心思想是使用在标记数据上学习的模型为未标记的样本生成伪标签,并反过来自我学习。为了获得有效的监督,主动尝试通常使用动量老师进行伪标签预测,但观察确认偏差问题,其中不正确的预测可能提供错误的监督信号并在训练过程中积累。这种缺陷的主要原因是,流行的自我训练框架充当了用以前的知识指导当前状态的角色,因为老师只更新了过去的学生。为了缓解这个问题,我们提出了一种新的自我训练策略,它允许模型从未来学习。具体地说,在每个训练步骤中,我们首先虚拟地优化学生(即,缓存梯度而不将它们应用于模型权重),然后用虚拟的未来学生更新教师,最后要求教师为当前学生产生伪标签作为指导。通过这种方式,我们设法提高伪标签的质量,从而提高性能。我们还开发了我们的未来自我训练(FST)框架的两个变体,通过深入(FST-D)和广泛(FST-W)窥视未来。以无监督的领域自适应语义分割和半监督的语义分割为例,实验证明了该方法在大范围场景下的有效性和优越性.代码将公开提供。

英文摘要:

Self-training has shown great potential in semi-supervised learning. Its core idea is to use the model learned on labeled data to generate pseudo-labels for unlabeled samples, and in turn teach itself. To obtain valid supervision, active attempts typically employ a momentum teacher for pseudo-label prediction yet observe the confirmation bias issue, where the incorrect predictions may provide wrong supervision signals and get accumulated in the training process. The primary cause of such a drawback is that the prevailing self-training framework acts as guiding the current state with previous knowledge, because the teacher is updated with the past student only. To alleviate this problem, we propose a novel self-training strategy, which allows the model to learn from the future. Concretely, at each training step, we first virtually optimize the student (i.e., caching the gradients without applying them to the model weights), then update the teacher with the virtual future student, and finally ask the teacher to produce pseudo-labels for the current student as the guidance. In this way, we manage to improve the quality of pseudo-labels and thus boost the performance. We also develop two variants of our future-self-training (FST) framework through peeping at the future both deeply (FST-D) and widely (FST-W). Taking the tasks of unsupervised domain adaptive semantic segmentation and semi-supervised semantic segmentation as the instances, we experimentally demonstrate the effectiveness and superiority of our approach under a wide range of settings. Code will be made publicly available.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑