BriefGPT.xyz
Jan, 2021
全局凸优化:通过神经半定程序提升多项式激活神经网络的神经谱聚合
Neural Spectrahedra and Semidefinite Lifts: Global Convex Optimization of Polynomial Activation Neural Networks in Fully Polynomial-Time
HTML
PDF
Burak Bartan, Mert Pilanci
TL;DR
本文提出了基于半定规划的二层神经网络的精确凸优化公式,可在多种神经网络体系结构中实现全局最优解,相比标准反向传播方法速度更快且准确性更高。
Abstract
The training of
two-layer neural networks
with
nonlinear activation functions
is an important non-convex optimization problem with numerous applications and promising performance in layerwise deep learning. In th
→