Shahine Bouabid

• Personal Webpage
• GitHub
• Twitter
• Google Scholar

Kernel Methods, Gaussian processes, Climate emulation

I am interested in developing simple and interpretable statistical machine learning methodologies to address challenges that arise in climate science. My recent work has focused on statistical downscaling and earth system emulation using physically-constrained models. The tools I use mostly draw from the theory of reproducing kernel Hilbert spaces and Gaussian processes, for which I enjoy a fond theoretical interest.

I am currently in the 3rd year of my PhD, supervised by Dino Sejdinovic, Tom Rainforth and Athanasios Nenes as part of the iMiracli innovative training network of aerosols-cloud interactions and machine learning.

Publications

2023

• S. Bouabid , J. Fawkes , D. Sejdinovic , Returning The Favour: When Regression Benefits From Probabilistic Causal Knowledge, arXiv preprint arXiv:2301.11214, 2023.
  

  @article{bouabid2023returning,
    title = {Returning The Favour: When Regression Benefits From Probabilistic Causal Knowledge},
    author = {Bouabid, Shahine and Fawkes, Jake and Sejdinovic, Dino},
    journal = {arXiv preprint arXiv:2301.11214},
    year = {2023}
  }
  

2022

• S. Bouabid , D. Watson-Parris , D. Sejdinovic , Bayesian inference for aerosol vertical profiles, in NeurIPS Workshop on Tackling Climate Change with Machine Learning, 2022.
  

  @inproceedings{bouabid2022bayesian,
    title = {Bayesian inference for aerosol vertical profiles},
    author = {Bouabid, Shahine and Watson-Parris, Duncan and Sejdinovic, Dino},
    year = {2022},
    booktitle = {NeurIPS Workshop on Tackling Climate Change with Machine Learning}
  }
  

• S. Bouabid , D. Watson-Parris , S. Stefanović , A. Nenes , D. Sejdinovic , AODisaggregation: toward global aerosol vertical profiles, arXiv preprint arXiv:2205.04296, 2022.
  

  @article{bouabid2022aodisaggregation,
    title = {AODisaggregation: toward global aerosol vertical profiles},
    author = {Bouabid, Shahine and Watson-Parris, Duncan and Stefanovi{\'c}, Sofija and Nenes, Athanasios and Sejdinovic, Dino},
    journal = {arXiv preprint arXiv:2205.04296},
    year = {2022}
  }
  

• D. Watson-Parris , Y. Rao , D. Olivié , Ø. Seland , P. Nowack , G. Camps-Valls , P. Stier , S. Bouabid , M. Dewey , E. Fons , . others , ClimateBench v1. 0: A Benchmark for Data-Driven Climate Projections, Journal of Advances in Modeling Earth Systems, vol. 14, no. 10, e2021MS002954, 2022.
  

  @article{watson2022climatebench,
    title = {ClimateBench v1. 0: A Benchmark for Data-Driven Climate Projections},
    author = {Watson-Parris, Duncan and Rao, Yuhan and Olivi{\'e}, Dirk and Seland, {\O}yvind and Nowack, Peer and Camps-Valls, Gustau and Stier, Philip and Bouabid, Shahine and Dewey, Maura and Fons, Emilie and others},
    journal = {Journal of Advances in Modeling Earth Systems},
    volume = {14},
    number = {10},
    pages = {e2021MS002954},
    year = {2022},
    publisher = {Wiley Online Library}
  }
  

2021

• S. L. Chau , S. Bouabid , D. Sejdinovic , Deconditional Downscaling with Gaussian Processes, in Advances in Neural Information Processing Systems (NeurIPS), 2021.
  
  Refining low-resolution (LR) spatial fields with high-resolution (HR) information is challenging as the diversity of spatial datasets often prevents direct matching of observations. Yet, when LR samples are modeled as aggregate conditional means of HR samples with respect to a mediating variable that is globally observed, the recovery of the underlying fine-grained field can be framed as taking an "inverse" of the conditional expectation, namely a deconditioning problem. In this work, we introduce conditional mean processes (CMP), a new class of Gaussian Processes describing conditional means. By treating CMPs as inter-domain features of the underlying field, a posterior for the latent field can be established as a solution to the deconditioning problem. Furthermore, we show that this solution can be viewed as a two-staged vector-valued kernel ridge regressor and show that it has a minimax optimal convergence rate under mild assumptions. Lastly, we demonstrate its proficiency in a synthetic and a real-world atmospheric field downscaling problem, showing substantial improvements over existing methods.

  @inproceedings{ChaBouSej2021,
    title = {{Deconditional Downscaling with Gaussian Processes}},
    author = {Chau, Siu Lun and Bouabid, Shahine and Sejdinovic, Dino},
    year = {2021},
    booktitle = {Advances in Neural Information Processing Systems (NeurIPS)}
  }
  

2020

• P. Harder , W. Jones , R. Lguensat , S. Bouabid , J. Fulton , D. Quesada-Chacón , A. Marcolongo , S. Stefanović , Y. Rao , P. Manshausen , D. Watson-Parris , NightVision: Generating Nighttime Satellite Imagery from Infra-Red Observations, in NeurIPS Workshop on Tackling Climate Change with Machine Learning, 2020.
  

  @inproceedings{harder2020nightvision,
    title = {NightVision: Generating Nighttime Satellite Imagery from Infra-Red Observations},
    author = {Harder, Paula and Jones, William and Lguensat, Redouane and Bouabid, Shahine and Fulton, James and Quesada-Chacón, Dánell and Marcolongo, Aris and Stefanović, Sofija and Rao, Yuhan and Manshausen, Peter and Watson-Parris, Duncan},
    year = {2020},
    booktitle = {NeurIPS Workshop on Tackling Climate Change with Machine Learning}
  }
  

• S. Bouabid , M. Chernetskiy , M. Rischard , J. Gamper , Predicting Landsat Reflectance with Deep Generative Fusion, in NeurIPS Workshop on Tackling Climate Change with Machine Learning, 2020.
  

  @inproceedings{bouabid2020predicting,
    title = {Predicting Landsat Reflectance with Deep Generative Fusion},
    author = {Bouabid, Shahine and Chernetskiy, Maxim and Rischard, Maxime and Gamper, Jevgenij},
    year = {2020},
    booktitle = {NeurIPS Workshop on Tackling Climate Change with Machine Learning}
  }