论文标题
解剖神经氧
Dissecting Neural ODEs
论文作者
论文摘要
连续深度学习架构最近重新出现为神经普通微分方程（神经odes）。这种无限的深度方法理论上弥合了深度学习和动态系统之间的差距，从而提供了一种新颖的视角。但是，对这些模型的内部工作进行解密仍然是一个开放的挑战，因为大多数应用程序将它们用作通用黑盒模块。在这项工作中，我们“打开盒子”，进一步开发连续的深度配方，目的是阐明几种设计选择对基础动力学的影响。
Continuous deep learning architectures have recently re-emerged as Neural Ordinary Differential Equations (Neural ODEs). This infinite-depth approach theoretically bridges the gap between deep learning and dynamical systems, offering a novel perspective. However, deciphering the inner working of these models is still an open challenge, as most applications apply them as generic black-box modules. In this work we "open the box", further developing the continuous-depth formulation with the aim of clarifying the influence of several design choices on the underlying dynamics.
