Björn Lütjens
@bjornlutjens.bsky.social
68 followers 64 following 5 posts
Research Scientist@IBM (he/him). Previously MIT. Research in machine learning and Earth system modeling 🌎 https://blutjens.github.io/
Posts Media Videos Starter Packs
bjornlutjens.bsky.social
Thanks to collaborators Raffaele Ferrari, @dwatsonparris.bsky.social, and Noelle Selin, and @mit-cs3.bsky.social ‪‪@eapsmit.bsky.social‬ (Adam Zewe/MIT News)
bjornlutjens.bsky.social
MIT News writes today about our new paper: A simple linear model can outperform a large-scale AI foundation model in climate emulation. We offer an explanation with lessons on benchmarking, bias-variance tradeoff, and natural climate variability.

news.mit.edu/2025/simpler...
A table with NRMSE scores showing that Linear Regression has a better score than the Foundation Model, Vision Transformer, Random Forest, CNN-LSTM, and Gaussian Process.
bjornlutjens.bsky.social
AGU abstracts due today, 07/30, 23.59 EDT. We're organizing the session on Advances in Emulating Earth System Models with speakers Mike Pritchard (NVIDIA), Sarah Schöngart (Climate Analytics), and Noelle Selin (MIT).

Submission link:
agu.confex.com/agu/agu25/pr...
Advances in Emulating Earth System Models
Running Earth system models (ESMs) requires a heavy computational cost. As a result, ESMs are often only run for a few emission scenarios with a limited number of realizations per scenario. Emulators ...
agu.confex.com
bjornlutjens.bsky.social
We just released the first lecture from this year's workshop on Computer Vision for Ecology at Caltech with MIT Prof Sara Beery giving an Intro to Computer Vision - Deep Learning, CNNs, and Transformers. Pls lmk if you have any comments:) youtu.be/5rE2ogBh0zc?...
Caltech CV4E - Intro to Computer Vision - Deep Learning, CNNs, Transformers - Prof. Sara Beery
YouTube video by CV4Ecology
youtu.be