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 2008.02389cs.LGmath.DSstat.ML
收藏

深度连续神经网络

Continuous-in-Depth Neural Networks

Google Research(谷歌研究院);ICSI and UC Berkeley(国际计算机科学研究所和加州大学伯克利分校);University at Buffalo, SUNY(纽约州立大学布法罗分校)

Alejandro F. Queiruga, N. Benjamin Erichson, Dane Taylor, Michael W. Mahoney

摘要:

最近的工作试图将剩余网络(ResNets)解释为常微分方程的前向欧拉离散化的一个步骤,主要关注两个系统之间的语法代数相似性。然而,连续动力系统的离散动力积分器具有更丰富的结构。我们首先证明,ResNets在这个更丰富的意义上不能成为有意义的动态积分器。然后,我们证明了神经网络模型可以学习表示连续动力系统,具有更丰富的结构和属性,通过将它们嵌入高阶数值积分方案,如Runge Kutta方案。基于这些见解,我们引入ContinuousNet作为ResNet架构的连续深入概括。ContinuousNets表现出对特定计算图表现的不变性。也就是说,可以用不同的离散时间步长(这改变了层数)和不同的数值积分方案(这改变了图形连通性)来评估深度连续模型。我们表明,这可以用来开发一种增量深度训练方案,提高模型质量,同时显着减少训练时间。我们还表明,一旦经过训练,计算图中的单元数量甚至可以减少,以便在几乎没有准确性下降的情况下进行更快的推理。

英文摘要:

Recent work has attempted to interpret residual networks (ResNets) as one step of a forward Euler discretization of an ordinary differential equation, focusing mainly on syntactic algebraic similarities between the two systems. Discrete dynamical integrators of continuous dynamical systems, however, have a much richer structure. We first show that ResNets fail to be meaningful dynamical integrators in this richer sense. We then demonstrate that neural network models can learn to represent continuous dynamical systems, with this richer structure and properties, by embedding them into higher-order numerical integration schemes, such as the Runge Kutta schemes. Based on these insights, we introduce ContinuousNet as a continuous-in-depth generalization of ResNet architectures. ContinuousNets exhibit an invariance to the particular computational graph manifestation. That is, the continuous-in-depth model can be evaluated with different discrete time step sizes, which changes the number of layers, and different numerical integration schemes, which changes the graph connectivity. We show that this can be used to develop an incremental-in-depth training scheme that improves model quality, while significantly decreasing training time. We also show that, once trained, the number of units in the computational graph can even be decreased, for faster inference with little-to-no accuracy drop.

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。

↑