Publicly Available: Recording, Slides

Synopsis: Training and evaluation performance for deep neural network models depends on a variety of factors, including type/size/performance of data storage solutions and on the type/configuration of computational hardware. These issues are especially acute as models and/or data sets become very large. In this tutorial, we discuss the current OU supercomputer architecture and procedures for monitoring and optimizing execution performance of deep learning models. Topics include: use of one or more Graphical Processing Units (GPUs), creating data loading and transformation pipelines, caching data to high-speed storage, monitoring CPU/GPU utilization, and procedures for coordinating shared use of multiple GPUs on a single compute node. We assume prior knowledge of Python, machine learning and deep networks.

Presenters: Mel Wilson Reyes, Jay Rothenberger, Andrew Fagg (OU)

A 4-part virtual short course on transformers.

• Transformers Part I: Recurrent Neural Networks & the Gradient Problem; Solutions using Attention – Andrew H. Fagg. Recording, Slides
• Transformers Part II: Positional Encodings and the Basic Transformer Architecture – Andrew H. Fagg. Recording, Slides (for Sessions 1 and 2) 
• Transformers Part III: Transformers in 2D – Andrew H. Fagg. Recording, Slides (for Sessions 1-3)
• Transformers Application: Transformer for Meteorological Forecasting Based on Multi-dimensional Input Data – Hamid Kamangir. Recording, Slides

Publicly Available: Slides, GitHub

A gentle introduction to machine learning for Earth scientists. This short covers topics such as machine learning for science, foundational concepts of ML, decision trees, random forests, artificial neural networks, ethical use of AI in Earth science, implementation and assessment of neural networks, methods of explainable AI (XAI), and simple uncertainty quantification.

Instructor: Elizabeth Barnes, Colorado State University

Publicly Available: Website

Online tutorial and resources to get started with Machine Learning

A non-exhaustive list of resources to help environmental scientists get started with machine learning, including tutorials with code. 

Instructor: Tom Beucler, University of Lausanne

Publicly Available Resource Links: Course Information, interactive blog for students, slides, user personas, and GitHub.

Offered by AI2ES and NCAR with partners LEAP, NCAI, and the Radiant Earth Foundation. Includes interactive, machine learning Trust-a-thon with beginner and advanced tracks

Participants will gain an understanding of:

1. the foundations of trustworthiness for AI 
2. explanatory AI (XAI) and how explanations, physics, and robustness can help build trust in AI
3. the relationship between ethics and trustworthiness
4. how machine-learning systems have been developed for a range of environmental science applications

Six-part tutorial on risk communication fro trustworthy AI from AI2ES members and prominent, invited researchers in the field.

This tutorial begins by putting risk communication research into context and introducing basics of risk perception and communication. It includes sessions on the role of trust in risk communication, and the design of risk communications and messages, and two related site-wide talks by guest speakers who are conducting research on formalizing trust in AI, and on human-AI teaming, respectively. Discussions, polls, and exercises are included to promote active learning. The concluding session focuses on what we can learn from iconic risk communication studies and meta-analyses for two topics: warnings, and fear appeals.

Recorded Sessions and Discussions:

1. Intro to risk perception, risk communication. AI2ES Members. Recording; Slides
2. Prerequisites, Causes, and Goals of Human Trust in AI. Ana Marasović. Recording
3. Trust in risk communication. AI2ES Members. Recording; Slides
4. Relational trust in human-agent teaming. Erin Chiou. Recording; Slides
5. Risk communication and message design basics. AI2ES Members. Recording; Slides
6. Iconic studies in risk communication (e.g., warning effectiveness).  AI2ES Members. Recording; Slides
   

Publicly Available: Course Information and Recordings, GitHub

The short course consists of six 90-minute lectures, each accompanied by slides and a Colab notebook with open-source Python code. 

All Python examples will focus on weather prediction. Made in collaboration between Cooperative Institute for Research in the Atmosphere (CIRA) and AI2ES.

Publicly Available: Course Information, Slides and Recordings

The TAI4ES Summer School will focus on getting attendees up-to-speed on how to develop trustworthy AI for the Earth & environmental sciences.  Each morning, students will participate in lectures from leading researchers in trustworthy AI for environmental sciences.

Topics include Trustworthy AI for Environmental Science; Explainable, Robust, Physics-based AI; AI, Ethics, and Trust; Case Studies; R2O Tips and Tricks.

Hosted jointly by AI2ES and NCAR.

Publicly Available: Slides and Recordings

The AI4ESS Summer School is a week long interactive course covering essential and cutting edge topics in artificial intelligence, machine learning, and deep learning that are relevant for Earth System Science problems. Host: National Center for Atmospheric Research (Boulder, Colorado, USA)

Machine Learning in Python for Environmental Science Problems

April 8 – 9, 2021

Hosted by American Meteorological Society, co-developed by AI2ES

Interest in artificial intelligence, machine learning, and deep learning in the environmental sciences has grown rapidly in conjunction with the increased presence of these technologies in our daily lives. Many people now want to apply machine learning to their own data and problems but do not know where to start. This 2-day virtual short course will develop your skills in applying Python libraries to process your data, train a variety of machine-learning models, generate predictions, and evaluate and interpret your trained models for physical understanding.

Course Information 

2nd NOAA Workshop On Leveraging AI In Environmental Sciences

July 30, 2020 – February 25, 2021

Host: National Oceanic and Atmospheric Administration’s STAR Center for Satellite Applications and Research, co-developed by A2ES members.

The workshop will build on the success of the 1^st Workshop on Leveraging AI. This workshop will help gather scientists, program managers, and leaders from public, academic and private sectors. It will enable experts involved in the development and adaptation of AI tools and applications in a number of fields to meet and exchange experiences.

Workshop includes presentations, posters, panel discussions, and tutorials. 

Resource Links: Tutorial sessions, Workshop Information