Zobrazeno 1 - 10
of 54
pro vyhledávání: '"Chen, Zhengdao"'
Autor:
Chen, Zhengdao
Publikováno v:
Journal of Machine Learning Research, 25(109):1-65, 2024
To characterize the function space explored by neural networks (NNs) is an important aspect of learning theory. In this work, noticing that a multi-layer NN generates implicitly a hierarchy of reproducing kernel Hilbert spaces (RKHSs) - named a neura
Externí odkaz:
http://arxiv.org/abs/2307.01177
Oversmoothing is a central challenge of building more powerful Graph Neural Networks (GNNs). While previous works have only demonstrated that oversmoothing is inevitable when the number of graph convolutions tends to infinity, in this paper, we preci
Externí odkaz:
http://arxiv.org/abs/2212.10701
To understand the training dynamics of neural networks (NNs), prior studies have considered the infinite-width mean-field (MF) limit of two-layer NN, establishing theoretical guarantees of its convergence under gradient flow training as well as its a
Externí odkaz:
http://arxiv.org/abs/2210.16286
We study the optimization of wide neural networks (NNs) via gradient flow (GF) in setups that allow feature learning while admitting non-asymptotic global convergence guarantees. First, for wide shallow NNs under the mean-field scaling and with a gen
Externí odkaz:
http://arxiv.org/abs/2204.10782
From the perspective of expressive power, this work compares multi-layer Graph Neural Networks (GNNs) with a simplified alternative that we call Graph-Augmented Multi-Layer Perceptrons (GA-MLPs), which first augments node features with certain multi-
Externí odkaz:
http://arxiv.org/abs/2010.15116
Recent theoretical works have characterized the dynamics of wide shallow neural networks trained via gradient descent in an asymptotic mean-field limit when the width tends towards infinity. At initialization, the random sampling of the parameters le
Externí odkaz:
http://arxiv.org/abs/2008.09623
Publikováno v:
34th Conference on Neural Information Processing Systems (NeurIPS 2020)
The ability to detect and count certain substructures in graphs is important for solving many tasks on graph-structured data, especially in the contexts of computational chemistry and biology as well as social network analysis. Inspired by this, we p
Externí odkaz:
http://arxiv.org/abs/2002.04025
Publikováno v:
8th International Conference on Learning Representations (ICLR 2020)
We propose Symplectic Recurrent Neural Networks (SRNNs) as learning algorithms that capture the dynamics of physical systems from observed trajectories. An SRNN models the Hamiltonian function of the system by a neural network and furthermore leverag
Externí odkaz:
http://arxiv.org/abs/1909.13334
Publikováno v:
Original version published at the 33rd Conference on Neural Information Processing Systems (NeurIPS 2019)
Graph Neural Networks (GNNs) have achieved much success on graph-structured data. In light of this, there have been increasing interests in studying their expressive power. One line of work studies the capability of GNNs to approximate permutation-in
Externí odkaz:
http://arxiv.org/abs/1905.12560
Publikováno v:
E3S Web of Conferences, Vol 553, p 01006 (2024)
Lithium batteries, as the most closely watched secondary batteries, exhibit outstanding performance. Nonetheless, the formation of lithium dendrites significantly compromises their safety and hampers further development. To mitigate the impact of lit
Externí odkaz:
https://doaj.org/article/99895279faf945a381be083437c6df7e