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 2104.12229cs.CV
收藏

向量神经元:SO的一般框架(3)-等变网络

Vector Neurons: A General Framework for SO(3)-Equivariant Networks

Stanford University(斯坦福大学);NVIDIA(英伟达);Google Research(谷歌研究院);University of Toronto(多伦多大学)

Congyue Deng, Or Litany, Yueqi Duan, Adrien Poulenard, Andrea Tagliasacchi, Leonidas Guibas

摘要:

旋转组的不变性和等变性在点云的3D深度学习社区中得到了广泛讨论。然而,大多数提出的方法要么使用复杂的数学工具,可能会限制其可访问性,或绑定到特定的输入数据类型和网络架构。在本文中,我们介绍了一个通用的框架,我们称之为矢量神经元表示创建SO(3)-等变神经网络点云处理。将神经元从1D标量扩展到3D向量,我们的向量神经元能够将SO(3)动作简单映射到潜在空间,从而为在常见的神经操作中构建等方差提供了框架-包括线性层,非线性,池化和归一化。由于它们的简单性,向量神经元是通用的,正如我们所展示的,可以被纳入不同的网络架构骨干,使它们能够处理任意姿势的几何输入。尽管它的简单性,我们的方法在准确性和概括性方面与其他更复杂和专业的最先进的分类和分割任务的方法相比表现得非常好。我们还首次展示了一个旋转等变重构网络。

英文摘要:

Invariance and equivariance to the rotation group have been widely discussed in the 3D deep learning community for pointclouds. Yet most proposed methods either use complex mathematical tools that may limit their accessibility, or are tied to specific input data types and network architectures. In this paper, we introduce a general framework built on top of what we call Vector Neuron representations for creating SO(3)-equivariant neural networks for pointcloud processing. Extending neurons from 1D scalars to 3D vectors, our vector neurons enable a simple mapping of SO(3) actions to latent spaces thereby providing a framework for building equivariance in common neural operations -- including linear layers, non-linearities, pooling, and normalizations. Due to their simplicity, vector neurons are versatile and, as we demonstrate, can be incorporated into diverse network architecture backbones, allowing them to process geometry inputs in arbitrary poses. Despite its simplicity, our method performs comparably well in accuracy and generalization with other more complex and specialized state-of-the-art methods on classification and segmentation tasks. We also show for the first time a rotation equivariant reconstruction network.

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。

↑