arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1902.02880cs.LGcond-mat.dis-nncond-mat.stat-mechstat.ML
收藏

多层神经网络学习动力学的平均场极限

Mean Field Limit of the Learning Dynamics of Multilayer Neural Networks

Stanford University(斯坦福大学)

Phan-Minh Nguyen

摘要:

多层神经网络--通常被构造成具有许多跨层非线性激活神经元的高度复杂结构--是否可以以一种非平凡的方式表现,但简化了其复杂性的主要部分?在这项工作中,我们发现了一种现象,在这种现象中,这些复杂网络的行为-在适当的缩放和随机梯度下降动力学-成为独立的神经元的数量,因为这个数字增长足够大。我们开发了一种形式主义,其中许多神经元的限制行为被捕获的一组方程,从而暴露了这些网络的一个以前未知的操作制度。虽然目前的追求在数学上是不严格的,但它补充了几个实验来验证这种行为的存在。

英文摘要:

Can multilayer neural networks -- typically constructed as highly complex structures with many nonlinearly activated neurons across layers -- behave in a non-trivial way that yet simplifies away a major part of their complexities? In this work, we uncover a phenomenon in which the behavior of these complex networks -- under suitable scalings and stochastic gradient descent dynamics -- becomes independent of the number of neurons as this number grows sufficiently large. We develop a formalism in which this many-neurons limiting behavior is captured by a set of equations, thereby exposing a previously unknown operating regime of these networks. While the current pursuit is mathematically non-rigorous, it is complemented with several experiments that validate the existence of this behavior.

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。

↑