UW-Atmos-MLJC

Machine-Learning Journal Club Meeting Collection

Website Version

GitHub Version

Department of Atmospheric and Climate Science at the University of Washington

This is the collection of all the MLJC meeting materials.

---

MLJC Meetings

Resources

---

Resources

Department of Atmospheric and Climate Science (Just in case you need it)

Topics People Interested

• Favorite ML topics:
  • Math behind ML (linalg, stats, theory) x5
  • Satellite x2, Observations
  • Linear Systems, DMD
  • Operational forecast
  • Data Assimilation
  • Emulator
  • Explainable AI
  • ML for improving science, Capability of ML model (speed, backprop)
  • Generative model
  • NOT math
  • NOT training
• ML topics people wish to know more about:
  • Generative model x2
  • Climate and AI x2
  • ML methods x2
  • Benchmarking (evaluation) x3, Uncertainty quantification
  • Process/Dynamics emulation
  • Reduced Order Models
  • Sampling
  • PDE problems
  • NOT ML Architecture

Tutorial

• A Recipe for Training Neural Networks
• Learning Machine Learning with Lorenz-96
• Statistical Machine Learning with R
  • This tutorial is written in R, but they provide many mathematical derivations of basic ML building blocks. No Neural Networks.
• Neural Network Playground
  • Experimenting with NN classifications.
• BST236 Computing Class from Harvard
• Zarr Introduction

Specific Topics

• The most complex model we actually understand

Library, Documentation

• scikit-learn
  • User Guide
  • Docs
• PyTorch
  • Python Docs
• TensorFlow
  • Learn
  • Python Docs

Please email Andy and Yi any suggestions.

Paper for Future Meetings

• Incorporating Multivariate Consistency in ML-Based Weather Forecasting with Latent-Space Constraints
• Evaluating machine learning weather models for data assimilation: fundamental limitations in tangent linear and adjoint properties
• Deep Learning Parameterization of Vertical Wind Velocity Variability via Constrained Adversarial Training
• Deep learning representation of the aerosol size distribution
• Probabilistic Precipitation Nowcasting with Rectified Flow Transformers
• Short-Term Hourly Weather Forecasting Using PredRNN With Image Preprocessing

Please email Andy and Yi any suggestions.

This site is open source. Improve this page.