The Aim

Over the past few years machine learning has emerged as an important tool for scientists, being utilized for a range of tasks including improving model parameterizations, identifying extreme weather events, improving predictions across timescales, clustering data samples and general non-linear data analysis. We would like to provide a multi-day machine learning tutorial for atmospheric scientists at CSU. 

This tutorial assumes no prior knowledge of machine learning. 

• Day 1: start from the basics (e.g. back propagation, training and testing) and basic architectures (e.g. neural networks, random forest, autoencoders, convolutional neural networks, etc.) 

• Day 2: build on day 1's discussion by digging deeper into implementation and visualization techniques (e.g. running actual code, optimal input, layer-wise relevance propagation)

• Optional Day 3: students bring their lunches and laptops+data to receive guidance applying what they've learned to their own research

Schedule

Day One: The Basics of Machine Learning

1:00pm - 1:15pm — The Big Idea of Machine Learning for Science (Libby)

1:15pm - 1:35pm — Foundational Concepts and Machine Learning Approach (Wei-Ting and Jamin)

1:35pm - 2:25pm — Decision Trees and Random Forest (Aaron)

2:25pm - 2:45pm — Break

2:45pm - 4:00pm — Neural Networks (Libby and Jamin)

4:00pm - 4:35pm — Advanced Topics related to Neural Networks (Yoonjin and Kirsten)

4:35pm - 5:00pm — Machine Learning Research in ATS

Day Two: Implementation and Visualization

1:00pm - 1:15pm — Machine Learning Research in ATS (Libby)

1:15pm - 1:45pm — Forecasting Example using Random Forest (Aaron and Wei-Ting)

1:45pm - 2:20pm — Sine-wave and ENSO Examples using Neural Networks (Jamin and Ben)

2:20pm - 2:25pm — Discussion of Software Options (Ben)

2:25pm - 2:35pm — Visualization and Attribution: Opening the "Black Box" (Libby)

2:35pm - 2:55pm — Random Forest Visualization Techniques (Aaron)

2:55pm - 3:10pm — Break

3:10pm - 3:35pm — Random Forest Visualization Example (Aaron and Wei-Ting)

3:35pm - 4:15pm — Artificial Neural Network Visualization Techniques (Kirsten and Ben)

4:15pm - 4:40pm — Artificial Neural Network Visualization Example (Kirsten and Ben)

4:40pm - 4:50pm — Incorporating Physics into Machine Learning (Ben and Libby)

4:50pm - 5:00pm — Thoughtful use of AI for Earth Science (Libby)