BriefGPT.xyz
Jun, 2020
AI Feynman 2.0: 基于图模块化的帕累托最优符号回归
AI Feynman 2.0: Pareto-optimal symbolic regression exploiting graph modularity
HTML
PDF
Silviu-Marian Udrescu, Andrew Tan, Jiahai Feng, Orisvaldo Neto, Tailin Wu...
TL;DR
本篇研究提供一种改进的符号回归方法，利用Pareto最优化原理寻求最适合数据的公式，同时从神经网络的梯度属性中发现广义对称性，通过正常流将其推广到只有样本分布的概率分布上，并采用统计假设检验加速鲁棒性暴力搜索。
Abstract
We present an improved method for
symbolic regression
that seeks to fit data to formulas that are
pareto-optimal
, in the sense of having the best accuracy for a given complexity. It improves on the previous state
→