Contrastive learning often relies on comparing positive anchor samples with multiple negative samples to perform Self-Supervised Learning (SSL). However, non-contrastive approaches like BYOL, SimSiam, and Barlow Twins achieve SSL without explicit negative samples. In this paper, we introduce a unified matrix information-theoretic framework that explains many contrastive and non-contrastive learning methods. We then propose a novel method Matrix-SSL based on matrix information theory. Experimental results reveal that Matrix-SSL significantly outperforms state-of-the-art methods on the ImageNet dataset under linear evaluation settings and on MS-COCO for transfer learning tasks. Specifically, when performing 100 epochs pre-training, our method outperforms SimCLR by 4.6%, and when performing transfer learning tasks on MS-COCO, our method outperforms previous SOTA methods such as MoCo v2 and BYOL up to 3.3% with only 400 epochs compared to 800 epochs pre-training. Code available at https://github.com/yifanzhang-pro/Matrix-SSL.

翻译：对比学习通常依赖将正锚定样本与多个负样本进行对比以实现自监督学习（SSL）。然而，诸如BYOL、SimSiam和Barlow Twins等非对比方法在没有显式负样本的情况下实现了SSL。在本文中，我们引入了一个统一的矩阵信息论框架，该框架解释了多种对比和非对比学习方法。随后，我们提出了一种基于矩阵信息论的新方法Matrix-SSL。实验结果表明，在线性评估设置下，Matrix-SSL在ImageNet数据集上显著优于现有最先进方法，并在MS-COCO迁移学习任务中表现更佳。具体而言，当进行100轮预训练时，我们的方法比SimCLR高出4.6%；而在MS-COCO上执行迁移学习任务时，与仅使用400轮预训练（相比800轮）的方法相比，我们的方法比先前的SOTA方法（如MoCo v2和BYOL）提升了高达3.3%。代码见https://github.com/yifanzhang-pro/Matrix-SSL。