首页 > 其他分享> > 2021 Yann LeCun 团队对模型泛化性能的插值和外推理论的探讨:Learning in High Dimension Always Amounts to Extrapolation
2021 Yann LeCun 团队对模型泛化性能的插值和外推理论的探讨:Learning in High Dimension Always Amounts to Extrapolation
作者:互联网
Learning in High Dimension Always Amounts to Extrapolation
论文地址:
https://arxiv.org/abs/2110.09485
主要工作:
插值和外推( interpolation and extrapolation)的概念是从深度学习到函数近似的各个领域的基础。当样本落在给定数据集凸包的边界内时,样本
标签:泛化,Amounts,Always,High,Extrapolation,Learning,Dimension 来源: https://blog.csdn.net/weixin_44936889/article/details/120884215