A fascinating deep dive into "Ethics and Explainability in Climate AI: From Theory to Practice" featuring David John Gagne, Galen McKinley, Elizabeth A. Barnes, PhD and Savannah Thais 👩🏼‍💻, and chaired by Viviana Acquaviva at #ClimateInformatics 2024!

My takeways and a summary 🧵from the Ethics/Explainability panel at #ClimateInformatics: 1. Build models with a focus on user base 2. Consider explainability <-> consistency <-> robustness <-> value and impact 3. Embrace the inherently political role of technology/science 1/n

Join us again this year at the 6th NOAA AI Workshop focusing on heat resilience that covers topics across different environmental systems and their impact on society and ecosystems!

Best aurora shot from Longmont. 3 second exposure really helped bring out the colors.

10 years ago #OTD, Bing Xu and Tianqi Chen quietly announced the birth of XGBoost on the Kaggle #higgsml competition forum. Since then, XGBoost has skyrocketed in popularity to become the top ML go-to tool! 1/4

📣 Our paper is out! We use an interpretable neural network to disentangle the relative contributions of the MJO and ENSO to subseasonal predictability of atmospheric circulation over the North Pacific! William E Chapman NSF NCAR Climate and Global Dynamics Laboratory agupubs.onlinelibrary.wiley.com/doi/10.1029/20…

Looking back over the Denver hailstorm from last Thursday and it looks like the worst hail came right after a storm merger from the NE on a SE moving storm. Bet we got some hail injection from one storm to the other.

This is a beautiful, beautiful post. maggieappleton.com/home-cooked-so…

Gm to everyone who bumped to numpy 2.0, how’s it going?

So I just plotted the square root of the ERA5 surface geopotential and there are some wild patterns over the ocean. Looks like variations between -10 to 10 m2/s2 with regular patterns. Any idea what's going on with this?

New @nature paper: nature.com/articles/s4158… NeuralGCM results (all are a "first"): 1) A differentiable hybrid atmospheric model 2) Competitive with ECMWF ensemble 3) Competitive with a GCRM in a year-long simulation 4) 40-year AMIP-like runs. Smaller bias than AMIP runs. 1/

🥳 Paper Day! Give a read to my 1st article in #climate science, "Ethics in Climate AI: From Theory to Practice"! We advocate for applying principles of #ethicalAI in scientific inquiry and propose strategies+resources. W/ Elizabeth A. Barnes, PhD, David John Gagne, Galen McKinley, Savannah Thais 👩🏼‍💻

CNN: VP Harris has selected MN gov. Tim Walz to be her running mate.

✨🎨🏰Super excited to share our new paper Ensemble everything everywhere: Multi-scale aggregation for adversarial robustness Inspired by biology we 1) get adversarial robustness + interpretability for free, 2) turn classifiers into generators & 3) design attacks on vLLMs 1/12

It seems GitHub pages is down. Confirmed on two unrelated repos. Not shown on GitHub Status yet.

Ouch now the whole of Github is down.

embarrassment driven development: when you want to show a friend what you're working on, you look at the project in a different light and realize that you still have a ways to go, and jump on those with a renewed sense of urgency.

Excited to see the fruit of our Aspen Machine Learning for Climate workshop be released!

If you weren't able to get your abstract payment into AMS before the deadline, email the program chair for your conference and AMS conference staff. You might also still be able to submit payment on the abstract after the deadline but editing title/author/text is locked.

Diffusion is the rising tide that eventually submerges all frequencies, high and low 🌊 Diffusion is the gradual decomposition into feature scales, fine and coarse 🗼 Diffusion is just spectral autoregression 🤷🌈