On the generalization of learning algorithms that do not converge
On the generalization of learning algorithms that do not converge
复制标题
关于不收敛学习算法的泛化
DOI:
10.48550/arxiv.2208.07951
复制
发表时间:
2022
期刊:
影响因子:
--
通讯作者:
S. Jegelka
中科院分区:
文献类型:
--
作者:
N. Chandramoorthy;Andreas Loukas;Khashayar Gatmiry;S. Jegelka
Generalization analyses of deep learning typically assume that the training converges to a fixed point. But, recent results indicate that in practice, the weights of deep neural networks optimized with stochastic gradient descent often oscillate indefinitely. To reduce this discrepancy between theory and practice, this paper focuses on the generalization of neural networks whose training dynamics do not necessarily converge to fixed points. Our main contribution is to propose a notion of statistical algorithmic stability (SAS) that extends classical algorithmic stability to non-convergent algorithms and to study its connection to generalization. This ergodic-theoretic approach leads to new insights when compared to the traditional optimization and learning theory perspectives. We prove that the stability of the time-asymptotic behavior of a learning algorithm relates to its generalization and empirically demonstrate how loss dynamics can provide clues to generalization performance. Our findings provide evidence that networks that"train stably generalize better"even when the training continues indefinitely and the weights do not converge.
登录
查看更多内容
DOI:
--
发表时间:
2020-02
期刊:
ArXiv
影响因子:
--
作者:
Lingkai Kong;Molei Tao
通讯作者:
Lingkai Kong;Molei Tao
DOI:
--
发表时间:
2019
期刊:
Proceedings of the Thirty-Second Conference on Learning Theory
影响因子:
--
作者:
Mei, Song;Misiakiewicz, Theodor;Montanari, Andrea
通讯作者:
Montanari, Andrea
影响因子:
2.5
作者:
Liu, Chaoyue;Zhu, Libin;Belkin, Mikhail
通讯作者:
Belkin, Mikhail
DOI:
--
发表时间:
2021-06
期刊:
ArXiv
影响因子:
--
作者:
Andreas Loukas;Marinos Poiitis;S. Jegelka
通讯作者:
Andreas Loukas;Marinos Poiitis;S. Jegelka
DOI:
--
发表时间:
2022
期刊:
The International Conference on Learning Representations
影响因子:
--
作者:
Wang, Yuqing;Chen, Minshuo;Zhao, Tuo;Tao, Molei
通讯作者:
Tao, Molei