1 code implementation • 15 Apr 2024 • Daniel Zhengyu Huang, Jiaoyang Huang, Zhengjiang Lin
Score-based generative models have emerged as a powerful approach for sampling high-dimensional probability distributions.
no code implementations • 5 Oct 2023 • Yifan Chen, Daniel Zhengyu Huang, Jiaoyang Huang, Sebastian Reich, Andrew M Stuart
Our third contribution is to study, and develop efficient algorithms based on Gaussian approximations of the gradient flows; this leads to an alternative to particle methods.
no code implementations • 4 Oct 2023 • Gerard Ben Arous, Reza Gheissari, Jiaoyang Huang, Aukosh Jagannath
We rigorously study the joint evolution of training dynamics via stochastic gradient descent (SGD) and the spectra of empirical Hessian and gradient matrices.
1 code implementation • 30 May 2023 • Kenji Kawaguchi, Zhun Deng, Xu Ji, Jiaoyang Huang
In this paper, we provide the first rigorous learning theory for justifying the benefit of information bottleneck in deep learning by mathematically relating information bottleneck to generalization errors.
1 code implementation • 21 Feb 2023 • Yifan Chen, Daniel Zhengyu Huang, Jiaoyang Huang, Sebastian Reich, Andrew M. Stuart
The flow in the Gaussian space may be understood as a Gaussian approximation of the flow.
1 code implementation • 26 Nov 2022 • Han Gao, Xu Han, Jiaoyang Huang, Jian-Xun Wang, Li-Ping Liu
Recently the Transformer structure has shown good performances in graph learning tasks.
no code implementations • 27 Jun 2022 • Kenji Kawaguchi, Zhun Deng, Kyle Luh, Jiaoyang Huang
This paper proves that robustness implies generalization via data-dependent generalization bounds.
no code implementations • NeurIPS 2021 • Clement Gehring, Kenji Kawaguchi, Jiaoyang Huang, Leslie Kaelbling
Estimating the per-state expected cumulative rewards is a critical aspect of reinforcement learning approaches, however the experience is obtained, but standard deep neural-network function-approximation methods are often inefficient in this setting.
Model-based Reinforcement Learning reinforcement-learning +1
no code implementations • 19 Oct 2021 • Gérard Ben Arous, Daniel Zhengyu Huang, Jiaoyang Huang
In this paper, we consider the singular values and singular vectors of low rank perturbations of large rectangular random matrices, in the regime the matrix is "long": we allow the number of rows (columns) to grow polynomially in the number of columns (rows).
no code implementations • 21 Feb 2021 • Daniel Z. Huang, Jiaoyang Huang
The unscented Kalman inversion (UKI) presented in [1] is a general derivative-free approach to solving the inverse problem.
Numerical Analysis Numerical Analysis Optimization and Control
no code implementations • 1 Feb 2021 • Jiaoyang Huang, Horng-Tzer Yau
Consider the normalized adjacency matrices of random $d$-regular graphs on $N$ vertices with fixed degree $d\geq3$.
Probability Mathematical Physics Combinatorics Mathematical Physics 60B20, 05C80
no code implementations • ICML 2020 • Zhun Deng, Hangfeng He, Jiaoyang Huang, Weijie J. Su
An acknowledged weakness of neural networks is their vulnerability to adversarial perturbations to the inputs.
no code implementations • ICML 2020 • Jiaoyang Huang, Horng-Tzer Yau
However, it was observed in [5] that there is a performance gap between the kernel regression using the limiting NTK and the deep neural networks.
no code implementations • 5 Aug 2019 • Kenji Kawaguchi, Jiaoyang Huang
The theory developed in this paper only requires the practical degrees of over-parameterization unlike previous theories.
no code implementations • 7 Apr 2019 • Kenji Kawaguchi, Jiaoyang Huang, Leslie Pack Kaelbling
Furthermore, as special cases of our general results, this article improves or complements several state-of-the-art theoretical results on deep neural networks, deep residual networks, and overparameterized deep neural networks with a unified proof technique and novel geometric insights.
no code implementations • 20 Nov 2018 • Kenji Kawaguchi, Jiaoyang Huang, Leslie Pack Kaelbling
In this paper, we analyze the effects of depth and width on the quality of local minima, without strong over-parameterization and simplification assumptions in the literature.