论文标题
从无监督视频预测的未知因素中解开物理动态
Disentangling Physical Dynamics from Unknown Factors for Unsupervised Video Prediction
论文作者
论文摘要
通过偏微分方程（PDE）描述的利用物理知识是改善无监督视频预测方法的一种吸引人的方式。由于物理学对描述通用视频的完整视觉内容过于限制，因此我们介绍了Phydnet，这是一种两分支深度架构，它明确地将PDE Dynamics从未知的互补信息中删除。第二个贡献是提出一个新的经常性物理细胞（Phycell），灵感来自数据同化技术，用于在潜在空间中执行PDE受限的预测。在四个各种数据集上进行的广泛实验表明，植物网络对最先进方法的表现能力。消融研究还强调了分解和PDE受限预测带来的重要增益。最后，我们表明Phydnet提出了有趣的功能，用于处理缺失的数据和长期预测。
Leveraging physical knowledge described by partial differential equations (PDEs) is an appealing way to improve unsupervised video prediction methods. Since physics is too restrictive for describing the full visual content of generic videos, we introduce PhyDNet, a two-branch deep architecture, which explicitly disentangles PDE dynamics from unknown complementary information. A second contribution is to propose a new recurrent physical cell (PhyCell), inspired from data assimilation techniques, for performing PDE-constrained prediction in latent space. Extensive experiments conducted on four various datasets show the ability of PhyDNet to outperform state-of-the-art methods. Ablation studies also highlight the important gain brought out by both disentanglement and PDE-constrained prediction. Finally, we show that PhyDNet presents interesting features for dealing with missing data and long-term forecasting.
