Csaba Toth
University of Oxford
Andrew Wiles Building
Radcliffe Observatory Quarter
Woodstock Road
Oxford
OX2 6GG
Learning to Forget: Bayesian Time Series Forecasting using Recurrent Sparse Spectrum Signature Gaussian Processes
Csaba Tóth, Masaki Adachi, Michael A. Osborne, Harald Oberhauser. AISTATS 2025.
arxiv.org:2412.19727
Random Fourier Signature Features
Csaba Tóth, Harald Oberhauser, Zoltán Szabó. SIMODS 2025.
arxiv.org:2311.12214
Capturing Graphs with Hypo-Elliptic Diffusions
Csaba Tóth, Darrick Lee, Celia Hacker, Harald Oberhauser. NeurIPS 2022.
arxiv.org:2205.14092
Seq2Tens: An Efficient Representation of Sequences by Low-Rank Tensor Projections
Csaba Tóth, Patric Bonnier, Harald Oberhauser. ICLR 2021.
arxiv.org:2006.07027
Bayesian Learning from Sequential Data using Gaussian Processes with Signature Covariances
Csaba Tóth, Harald Oberhauser. ICML 2020.
arxiv.org:1906.08215
Scholar Award (NeurIPS, New Orleans, 2022)
Outstanding Reviewer Award (NeurIPS, Virtual, 2021)
Mathematical Institute Award (Mathematical Institute, Oxford, 2019)
Distinction Award (Lincoln College, Oxford, 2018)
Time Series and Graphs, Rough Paths, Signatures, Tensors, Bayesian Inference, Kernel Methods, Deep Learning, Concentration of Measure
MT 2023: Prelims Computational Mathematics (Demonstrator, Intercollegiate)
MT 2020, 2021: Part A Linear Algebra (Tutor, St. Hugh's)
HT 2020: Part C Limit Theorems and Large Deviations in Probability (TA, Intercollegiate)
MT 2019: Part C Theories of Deep Learning (TA, Intercollegiate)