Reference. LatticeVision: Image to Image Networks for Modeling Non-Stationary Spatial Data
In many applications, we wish to fit a parametric statistical model to a small ensemble of spatially distributed random variables (‘fields’). However, parameter inference using maximum likelihood estimation (MLE) is computationally prohibitive, especially for large, non-stationary fields. Thus, many recent works train neural networks to estimate parameters given spatial fields as input, sidestepping MLE completely. In this work we focus on a popular class of parametric, spatially autoregressive (SAR) models. We make a simple yet impactful observation; because the SAR parameters can be arranged on a regular grid, both inputs (spatial fields) and outputs (model parameters) can be viewed as images. Using this insight, we demonstrate that image-to-image (I2I) networks enable faster and more accurate parameter estimation for a class of non-stationary SAR models with unprecedented complexity.
Cite
Cites 84 works (0 here)
External (84)
- Modeling Spatial Extremes Using Non‐Gaussian Spatial Autoregressive Models via Convolutional Neural Networks (2025)
- CAMulator: Fast Emulation of the Community Atmosphere Model (2025)
- Learning Non-Gaussian Spatial Distributions Via Bayesian Transport Maps with Parametric Shrinkage (2024)
- CoDiCast: Conditional Diffusion Model for Global Weather Prediction with Uncertainty Quantification (2024)
- Pushing the frontiers in climate modelling and analysis with machine learning (2024)
- The Impact of Internal Variability on Benchmarking Deep Learning Climate Emulators (2024)
- Normalizing Basis Functions: Approximate Stationary Models for Large Spatial Data (2024)
- Machine Learning for Climate Physics and Simulations (2024)
- Neural Methods for Amortized Inference (2024)
- ChaosBench: A Multi-Channel, Physics-Based Benchmark for Subseasonal-to-Seasonal Climate Prediction (2024)
- Probablistic Emulation of a Global Climate Model with Spherical DYffusion (2024)
- GenCast: Diffusion-based ensemble forecasting for medium-range weather (2023)
- Scaling transformer neural networks for skillful and reliable medium-range weather forecasting (2023)
- Learning skillful medium-range global weather forecasting (2023)
- Neural general circulation models for weather and climate (2023)
- ClimateSet: A Large-Scale Climate Model Dataset for Machine Learning (2023)
- Neural Bayes Estimators for Irregular Spatial Data Using Graph Neural Networks (2023)
- Efficient Stochastic Generators with Spherical Harmonic Transformation for High-Resolution Global Climate Simulations from CESM2-LENS2 (2023)
- ACE: A fast, skillful learned global atmospheric model for climate prediction (2023)
- The New Max Planck Institute Grand Ensemble With CMIP6 Forcing and High‐Frequency Model Output (2023)
- Channel Vision Transformers: An Image Is Worth C x 16 x 16 Words (2023)
- Accurate medium-range global weather forecasting with 3D neural networks (2023)
- SEEDS: Emulation of Weather Forecast Ensembles with Diffusion Models (2023)
- DYffusion: A Dynamics-informed Diffusion Model for Spatiotemporal Forecasting (2023)
- Neural likelihood surfaces for spatial processes with computationally intensive or intractable likelihoods (2023)
- Fast parameter estimation of generalized extreme value distribution using neural networks (2023)
- DiffESM: Conditional Emulation of Earth System Models with Diffusion Models (2023)
- ClimaX: A foundation model for weather and climate (2023)
- ClimSim: A large multi-scale dataset for hybrid physics-ML climate emulation (2023)
- “Fengwu: Pushing the skillful global medium-range weather forecast beyond 10 days lead” (2023)
- The importance of internal climate variability in climate impact projections (2022)
- ClimateBench v1.0: A Benchmark for Data‐Driven Climate Projections (2022)
- Likelihood-Free Parameter Estimation with Neural Bayes Estimators (2022)
- HorNet: Efficient High-Order Spatial Interactions with Recursive Gated Convolutions (2022)
- InstaFormer: Instance-Aware Image-to-Image Translation with Transformer (2022)
- FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators (2022)
- Forecasting Global Weather with Graph Neural Networks (2022)
- A ConvNet for the 2020s (2022)
- Fast Gaussian Process Estimation for Large-Scale In Situ Inference using Convolutional Neural Networks (2021)
- Swin Transformer V2: Scaling Up Capacity and Resolution (2021)
- The SPDE approach for Gaussian and non-Gaussian fields: 10 years and still running (2021)
- Modelling Big, Heterogeneous, Non-Gaussian Spatial and Spatio-Temporal Data using FRK (2021)
- Neural networks for parameter estimation in intractable models (2021)
- Swin-Unet: Unet-like Pure Transformer for Medical Image Segmentation (2021)
- RoFormer: Enhanced Transformer with Rotary Position Embedding (2021)
- Learning to Optimize: A Primer and A Benchmark (2021)
- Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions (2021)
- Physics-informed machine learning: case studies for weather and climate modelling (2021)
- TransUNet: Transformers Make Strong Encoders for Medical Image Segmentation (2021)
- Ubiquity of human-induced changes in climate variability (2021)
- Fast covariance parameter estimation of spatial Gaussian process models using neural networks (2020)
- Training data-efficient image transformers & distillation through attention (2020)
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale (2020)
- Insights from Earth system model initial-condition large ensembles and future prospects (2020)
- Modeling spatial data using local likelihood estimation and a Matérn to spatial autoregressive translation (2020)
- How large does a large ensemble need to be? (2020)
- Task-Agnostic Amortized Inference of Gaussian Process Hyperparameters (2020)
- PyTorch: An Imperative Style, High-Performance Deep Learning Library (2019)
- The frontier of simulation-based inference (2019)
- Handbook of Approximate Bayesian Computation (2018)
- Meta-Kriging: Scalable Bayesian Modeling and Inference for Massive Spatial Datasets (2018)
- Group Normalization (2018)
- Image Transformer (2018)
- Modeling and emulation of nonstationary Gaussian fields (2017)
- Decoupled Weight Decay Regularization (2017)
- A Case Study Competition Among Methods for Analyzing Large Spatial Data (2017)
- Deep Sets (2017)
- Gaussian Error Linear Units (GELUs) (2016)
- Statistics For Spatial Data (2016)
- Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization (2015)
- The Community Earth System Model (CESM) large ensemble project: a community resource for studying climate change in the presence of internal climate variability (2015)
- A Multi-Resolution Approximation for Massive Spatial Datasets (2015)
- U-Net: Convolutional Networks for Biomedical Image Segmentation (2015)
- A Multiresolution Gaussian Process Model for the Analysis of Large Spatial Datasets (2015)
- Statistical emulation of climate model projections based on precomputed GCM runs (2014)
- Dropout: a simple way to prevent neural networks from overfitting (2014)
- An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach (2011)
- A modeling approach for large spatial datasets (2008)
- Anisotropic Matérn correlation and spatial prediction using REML (2007)
- Spectral methods for nonstationary spatial processes (2002)
- Interpolation of Spatial Data: Some Theory for Kriging (1999)
- Documentation of a highly ENSO‐related sst region in the equatorial pacific: Research note (1997)
- ON STATIONARY PROCESSES IN THE PLANE (1954)
- “Learning to Optimize: Where Deep Learning Meets Optimization and Inverse Problems”