EEPS-DATA 1720 is a new course, so the schedule and topics are tentative and subject to change.
Course Intro: Tackling Climate Change with Machine Learning (pdf) [Brown login required]
Readings (optional)
Karpatne, et al. (2018). Machine learning for the geosciences: Challenges and opportunities. IEEE Transactions on Knowledge and Data Engineering. https://arxiv.org/abs/1711.04708.
Reichstein, et al. (2019). Deep learning and process understanding for data-driven Earth system science. Nature. https://doi.org/10.1038/s41586-019-0912-1. [available online via Brown library].
Bortnik & Camporeale (2021). Ten ways to apply machine learning in Earth and space sciences. Eos. https://doi.org/10.1029/2021EO160257.
McGovern, et al. (2022). Why we need to focus on developing ethical, responsible, and trustworthy artificial intelligence approaches for environmental science. Environmental Data Science. doi:10.1017/eds.2022.5.
Lecture Slides
Climate 101 (Part 1) [pdf] [Brown login required]
Resources for Climate Information
Lecture Slides
References
[Animation] Molecules Vibrate (UCAR Center for Science Education)
Explainer: Climate Models (MIT Climate Portal)
Q&A: Why clouds are still 'one of the biggest uncertainties' in climate change (Horizon: The EU Research & Innovation Magazine, Nov 2020)
Climate Grand Challenges (World Climate Research Programme)
Lecture Slides
Readings
Carey et al. (2020). Ten Simple Rules for Reading a Scientific Paper, PLoS Computational Biology.
Keshav (2007). How to read a paper [pdf], ACM SIGCOMM Computer Communication Review.
Rasp (2021). “If you have a hammer…”: Five questions to ask before starting a machine learning project. Blog post.
[optional] Pain (2016). How to (seriously) read a scientific paper, Science Careers.
[optional] Introduction to Machine Learning Problem Framing, Google Developers Course.
Lecture Slides
Resources
Kashinath et al. (2021). Physics-informed machine learning: Case studies for weather and climate modeling [Review paper]. Philosophical Transactions of the Royal Society A.
Harder et al. (2022). Generating physically-consistent high-resolution climate data with hard-constrained neural networks. arXiv:2208.05424.
PI-NN example with Code: So what is a physics-informed neural network (blog post by Ben Moseley)
Code: Harmonic Oscillator [Colab notebook]
Case studies in Physics-Informed ML (see Kashinath et al.)
Case study A: Conditional GANs to emulate numerical-hydro-climate models, see Manepalli et al. (2019).
Case study B: Enforcing conservation laws in neural networks for climate modeling, see Beucler et al. (2019).
Case study C: Physics-constrained GAN for super-resolution weather data, see Singh et al. (2019b).
Lecture Materials
Presentation Slides (by David & Maria Luisa)
Notes from presentation/discussion [here]
Reading
Yuval, O'Gorman and Hill (2021). Use of Neural Networks for Stable, Accurate and Physically Consistent Parameterization of Subgrid Atmospheric Processes with Good Performance at Reduced Precision. Geophysical Research Letters.
Code available on GitHub.
Lecture Materials & Resources
Presentation slides (by Jack & Andy)
Notes from presentation/discussion [here]
Resources on Fourier Neural Operators
Neural Operators & Fourier Neural Operators, blog posts by Zongyi Li
FNO paper: https://arxiv.org/abs/2010.08895
Reading
Pathak et al. (2022). FourCastNet: A Global Data-Driven High-Resolution Weather Model using Adaptive Fourier Neural Operators. arXiv:2202.11214.
Code available on GitHub.
Lecture Materials
Presentation slides (by Iris & Jed)
Reading
Blanchard et al. (2022). A Multi-Scale Deep Learning Framework for Projecting Weather Extremes. NeurIPS Workshop on Tackling Climate Change with Machine Learning.
Lecture Slides
Reading
Kuglitsch et al. (2022). AI for Disaster Risk Reduction: Opportunities, challenges and prospects. World Meteorological Organization (WMO) Bulletin.
Resources
The human cost of weather-related disasters 1995-2015. UN Office for Disaster Risk Reduction (2015).
WMO Atlas of Mortality and Economic Losses from Weather, Cliamte and Water Extremes (1970-2019). World Meteorological Organization (2021).
U.S. Billion-Dollar Weather and Climate Disasters. NOAA National Centers for Environmental Information (2023). [dataset]
More dangerous heat waves are on the way: See the Impact by Zip code. Washington Post. August 15, 2022.
The National Risk Index. FEMA.
Lecture Slides
Readings
McGovern et al. (2019). Making the Black Box More Transparent: Understanding the Physical Implications of Machine Learning. Bulletin of the American Meteorological Society.
[optional] Toms et. al. (2020). Physically Interpretable Neural Networks for the Geosciences: Applications to Earth System Variability. JAMES.
Resources
Great introductory (free online) book: Interpretable Machine Learning: A Guide for Making Black Box Models Explainable by Christoph Molnar
Olah et al. (2017). Feature Visualization: How neural networks build up their understanding of images.
Interactive Demo on Layer-wise Relevance Propagation (LRP)
A critical perspective on XAI: Rudin (2019). Stop Explaining Black Box ML Models for High Stakes Decisions and Use Interpretable Models Instead.
Lecture Materials & Resources
Presentation Slides (by Michelle & Livia)
original U-net paper (58k+ citations as of 03/23) and supporting material
Reading
Hilburn et al. (2021). Development and Interpretation of a Neural-Network-Based Synthetic Radar Reflectivity Estimator using GOES-R Satellite Observations. Journal of Applied Meteorology and Climatology.
Lecture Materials
Lecture slides: Crowdsourcing in Earthquake Hazards Powered by ML (by Qingkai Kong)
MyShake project website - includes links to download app on Google Play and App Store
A recording of Dr. Kong's lecture is available by request to members of the Brown community - contact instructor for access.
Readings
Kong et al. (2016). MyShake: A Smartphone Seismic Network for Earthquake Early Warning and Beyond. Science Advances.
Kong et al. (2019). Machine Learning Aspects of the MyShake Global Smartphone Seismic Network. Seismological Research Letters.
Related work
Chachra et al. (2022). Detecting damaged buildings using real-time crowdsourced images and transfer learning. Scientific Reports.
Ruan et al. (2022). Cross-platform analysis of public responses to the 2019 Ridgecrest earthquake dequence on Twitter and Reddit. Scientific Reports.
Lecture Materials
Presentation slides (by Cali & John F.)
Resources
Google-produced videos: When Rivers Rise: How AI is helping predict floods & How AI is Helping Flood Forecasting | Inventors at Google
Readings
Nevo (2019). An Inside Look at Flood Forecasting. Google Research blog.
Nevo (2020a). The Technology Behind our Recent Improvements in Flood Forecasting. Google Research blog.
Nevo et al. (2020b). ML-based Flood Forecasting: Advances in Scale, Accuracy and Reach. AI for HADR Workshop, NeurIPS.
Moshe et al. (2020). HydroNets: Leveraging River Structure for Hydrological Modeling. AI for Physical Sciences Workshop, ICLR.
Related Work
Kratzert et al. (2019). Towards Learning Universal, Regional and Local Hydrological Behaviors via Machine learning Applied to Large-Sample Datasets. Hydrological and Earth Systems Science. [data and code available on GitHub]
Kratzert el al. (2019). Towards Improved Predictions in Ungauged Basins: Exploiting the Power of Machine Learning. Water Resources Research. [data and code available on GitHub]
Ben-Haim et al. (2019). Inundation Modeling in Data Scarce Regions. AI for HADR Workshop, NeurIPS.
Nevo et al. (2019). ML for Flood Forecasting at Scale. AI for Social Good Workshop, NeurIPS.
Slides (Google Slides)
Students are encouraged to attend the DEEPS Colloquium @ 12pm in MacMillan 115
Lecture Slides
Resources
Global Forest Change, Google Earth Engine App
IPCC, 2022. Climate Change 2022: Mitigation of Climate Change. Chapter 7: Agriculture, Forestry, and Other Land Uses.
IPCC, 2019: Climate Change and Land.
Kalaitzis et al. (2022). White Paper: State of AI for Earth Observations.
Kattenborn et al. (2021). Review on Convolutional Neural Networks in Vegetation Remote Sensing. ISPRS Journal of Photogrammetry and Remote Sensing.
Readings
Rolnick et al. (2022). Tackling Climate Change with ML - Section 6: Farms & Forests. ACM Computing Surveys.
Finer et al. (2018). Combating deforestation: From satellite to intervention. Science.
Lecture Materials
Lecture Slides: Integrating ML into Operational Weather Forecasts at the ECMWF (by Jesper Dramsch)
Reading
Düben et al. (2021). Machine Learning at ECMWF: A roadmap for the next 10 years. ECMWF Technical Memoranda.
ML Roadmap Webinar
Related Readings
Ben-Bouallegue et al. (2023). Improving medium-range ensemble weather forecasts with hierarchical ensemble transformers. (arXiv preprint)
Clare et al. (2022). Explainable AI for Bayesian Neural Networks: Toward Trustworthy Predictions of Ocean Dynamics. JAMES.
Chantry et al. (2021). Machine Learning Emulation of Gravity Wave Drag in Numerical Weather Forecasting. JAMES.
Data-driven weather forecasting papers:
FourCastNet (Pathak et al., 2022 / NVIDIA)
Pangu-Weather (Bi et al., 2022 / Huawei)
GraphCast (Lam et al., 2022 / DeepMind)
Resources
Projects
Educational Materials
Machine Learning in Weather & Climate MOOC
Software, Jupyter Notebooks, and Examples:
Inferno Library: a lower-level API for ML inference in operations
Copernicus Climate Change Service (C3S) Data Tutorials
Lecture Slides
Resources
Pan & Yang (2009). A Survey on Transfer Learning. IEEE Transactions on Knowledge and Data Engineering. [link to author pdf]
Finn et al. (2017). Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks. ICML.
Wang et al. (2019). Generalizing from a Few Examples: A Survey on Few-Shot Learning. ACM Computing Surveys.
Snell et al. (2017). Prototypical Networks for Few-Shot Learning. NeurIPS.
Van Engelen & Hoos (2020). A survey on semi-supervised learning. Machine Learning.
Reading
Xie et al. (2015). Transfer Learning from Deep Features for Remote Sensing and Poverty Mapping. AAAI.
Lectures
Presentation Slides (by Zain & Mason)
Resources
ClimaX: a foundation model for weather and climate. [website] [paper] [GitHub]
SatMAE: Pre-training Transformers for Temporal and Multi-spectral Satellite Imagery. [website] [paper] [GitHub]
Readings
Wang et al. (2022). Unlocking Large-Scale Crop Field Delineation in Smallholder Farming Systems with Transfer Learning and Weak Supervision. Remote Sensing.
[optional - see Canvas for instructions] Wang et al. (2020). Weakly Supervised Deep Learning for Segmentation of Remote Sensing Imagery. Remote Sensing. [code available on GitHub]
Lecture Slides
Presentation slides (by Panos, Sanyu & John N.)
Resources
An Interactive Introduction to Model-Agnostic Meta-Learning by Müller et al. (2021).
Hospedales et al. (2022). Meta-Learning in Neural Networks: A Survey. IEEE Transactions on Pattern Analysis and Machine Intelligence.
Meta-Learning: Learning to Learn Fast. Blog post by Lilian Weng (2018).
Learning to Learn. Blog post by Chelsea Finn (2017).
Tutorial 16: Meta-Learning: Learning to Learn. UvA Deep Learning Tutorials by Phillip Lippe.
Finn et al. (2017). Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks. ICML. [GitHub: Supervised & RL domains]
Readings
Rußwurm et al. (2020). Meta-learning for few-shot land cover classification. CVPRW.
Tseng et al. (2021). Learning to predict crop type from heterogeneous sparse labels using meta-learning. CVPRW. [GitHub]
Lecture Slides
Presentation slides (by Kai, Will & Joseph)
Resources
Tuia et al. (2022). Perspectives in machine learning for wildlife conservation. Nature.
Overview of Active Learning for Deep Learning. Blog post by Jacob Gildenblat.
Settles (2009). Active Learning Literature Survey. Technical Report - UW-Madison.
Readings
Norouzzadeh et al. (2020). A deep active learning system for species identification and counting in camera trap images. Methods in Ecology and Evolution.
Kellenberger et al. (2018). Detecting mammals in UAV images: Best practices to address a substantially imbalanced dataset with deep learning. Remote Sensing of the Environment.
Lecture Slides
Readings
McGovern et al. (2022). Why we need to focus on developing ethical, responsible, and trustworthy AI approached for environmental science. Environmental Data Science.
[optional] Coeckelbergh (2020). AI for climate: freedom, justice, and other ethical and political challenges. AI Ethics.
Google Slides [link]
Readings (see "Assignments" page for full readings for each case study)
Case Study A: CUREE: Curious Robot for Ecosystem Exploration. Project website.
Case Study B: Acoustic Detection of Humpback Whales Using a Convolutional Neural Network. Google Research Blog.
Case Study C: Coral Reef Video Game Will Help Create Global Database. Eos.
EEPS-DATA 1720 Final Presentations [Google slides] [pdf]