Houman Owhadi Group @ Caltech
@OwhadiGroup
Followers
105
Following
27
Statuses
44
Prof. @HoumanOwhadi's group @Caltech focuses on interplays between Numerical Approximation, Gaussian Processes and Learning. Tweets by students.
Pasadena, CA
Joined March 2022
Our latest paper on discovering functional relations using Gaussian processes is out in the Proceedings of the National Academy of Sciences @PNASNews @theNASciences! Read it at
1
2
4
RT @OwhadiGroup: Work by @TheoBourdais, @PauBatlle, Xianjin Yang, Ricardo Baptista, Nicolas Rouquette, and @HoumanOwhadi!
0
1
0
The entire source code is available as an installable Python package
0
2
1
RT @StatMLPapers: Computational Hypergraph Discovery, a Gaussian Process framework for connecting the dots. (arXiv:2311.17007v1 [cs.LG]) ht…
0
2
0
Our paper on kernel methods for operator learning is now out in the Journal of Computational Physics! We show that kernel methods match or beat neural operators for many PDE operator learning problems. Read it at
0
3
6
We present and analyze a method for frequentist confidence intervals in constrained inverse problems, refuting a conjecture from 1965 about coverage of gaussian constrained intervals! @PauBatlle @mstanley1123 Pratik Patil, Mikael Kuusala, @HoumanOwhadi
0
1
7
A very handy guide to navigate @iciam2023 starting today!
For anyone else who is struggling to figure out which ICIAM minisymposia to go to just by the code names, I did some parsing with @PauBatlle and made a sheet with a list of titles by timeslot here: (click on the bottom tab) #ICIAM2023
0
0
2
@StatMLPapers Our latest paper on analyzing kernel/GP methods for nonlinear and parametric PDEs! By @PauBatlle @yifanc96 @BamdadHosseini @HoumanOwhadi and Andrew Stuart
0
0
5
RT @StatMLPapers: Error Analysis of Kernel/GP Methods for Nonlinear and Parametric PDEs. (arXiv:2305.04962v1 [) htt…
0
2
0
RT @Matt_D_Darcy: New paper available on arxiv We present a simple kernel method for data-driven operator learni…
0
1
0
Generalization of the work where there is no derivative information
0
0
1
Our latest paper on kernel methods for operator learning is now available. We show that a simple kernel method is fast and accurate on many PDE operator learning problems, and competitive with DeepONet and FNO. With @PauBatlle @Matt_D_Darcy @BamdadHosseini @HoumanOwhadi
Kernel Methods are Competitive for Operator Learning. (arXiv:2304.13202v1 [)
0
1
7