no code implementations • 28 Jan 2022 • Takashi Mori, Masahito Ueda
It has been recognized that heavily overparameterized deep neural networks (DNNs) exhibit surprisingly good generalization performance in various machine-learning tasks.
no code implementations • 29 Sep 2021 • Takashi Mori, Liu Ziyin, Kangqiao Liu, Masahito Ueda
Stochastic gradient descent (SGD) undergoes complicated multiplicative noise for the mean-square loss.
no code implementations • 20 May 2021 • Takashi Mori, Liu Ziyin, Kangqiao Liu, Masahito Ueda
Stochastic gradient descent (SGD) undergoes complicated multiplicative noise for the mean-square loss.
no code implementations • 11 Feb 2021 • Takashi Mori
Metastable states in stochastic systems are often characterized by the presence of small eigenvalues in the generator of the stochastic dynamics.
Statistical Mechanics Disordered Systems and Neural Networks
no code implementations • ICLR 2022 • Liu Ziyin, Kangqiao Liu, Takashi Mori, Masahito Ueda
The noise in stochastic gradient descent (SGD), caused by minibatch sampling, is poorly understood despite its practical importance in deep learning.
no code implementations • 18 Jan 2021 • Takashi Mori, Hongzheng Zhao, Florian Mintert, Johannes Knolle, Roderich Moessner
The nonequilibrium quantum dynamics of closed many-body systems is a rich yet challenging field.
Statistical Mechanics Quantum Physics
no code implementations • 28 Sep 2020 • Takashi Mori, Masahito Ueda
Recent studies have demonstrated that noise in stochastic gradient descent (SGD) is closely related to generalization: A larger SGD noise, if not too large, results in better generalization.
no code implementations • 26 May 2020 • Takashi Mori, Masahito Ueda
It is shown that the NTK does not correctly capture the depth dependence of the generalization performance, which indicates the importance of the feature learning rather than the lazy learning.