BriefGPT.xyz
Jun, 2020
数据增强优化的元方法
Meta Approach to Data Augmentation Optimization
HTML
PDF
Ryuichiro Hataya, Jan Zdenek, Kazuki Yoshizoe, Hideki Nakayama
TL;DR
本文提出了一种优化图像识别模型和数据增强策略的方法,使用梯度下降同时优化两者,通过使用Neumann级数逼近来近似策略梯度,以实现高效可扩展的训练,以提高各种图像分类任务的性能。
Abstract
data augmentation
policies drastically improve the performance of
image recognition
tasks, especially when the policies are optimized for the target data and tasks. In this paper, we propose to optimize
→