Mackenzie Weygandt Mathis, PhD

🦓 Self-supervised multimodal ML is promising the next AI breakthrough - in our new work published in @Nature, we debut @CEBRAai: for self-supervised hypothesis- and discovery-driven science. 📝 doi.org/10.1038/s41586… 💻github.com/AdaptiveMotorC… 🦓 cebra.ai 🧵⬇️

From the neuroscience perspective, behavior is unmatched as noninvasive quantitative readout of brain function (and best of all, it is a readout that matters). This initial lifelong continuous screen of behaving vertebrates provided a wealth of new ideas regarding progression and staging of brain function across the adult lifespan. Truly impressive work from Claire, Ravi, and the whole creative and dedicated team. Lifelong work takes a long time, even with killifish!

💉 Finding clinical trials & knowing which are useful is an overwhelming task, even if you know where to start! I worked on a simple @Gradio app for parsing clinical trials+ adding a #DeepSeek model to rank trials + user Q&A. I hope its useful to many: clinicaltrialmatcher.org

How does the brain know which neurons to adjust during learning in order to optimize behavior? MIT researchers discovered that brains can use cell-by-cell error signals to do this — surprisingly similar to how AI systems are trained via backpropagation. mcgovern.mit.edu/2026/02/25/neu…

Swiss AI Visiting PhD Program at EPFL. Deadline Feb 28. "The program provides a fellowship contribution of CHF 2,500 ($3200) per month, access to the Alps supercomputer, and eligibility for a post-visit continuation grant of up to 50k GPU hours. The call is open to PhD students enrolled outside Switzerland, with applications supported by EPFL PIs contributing to the Swiss AI Initiative. The application closes on February 28." drive.google.com/file/d/1e0IEay… @ICepfl @EPFL_AI_Center

And now, accepted to #CVPR2026! 🙌🔥 @CVPR ⬇️

People often assume that state of the art vision models are *not recurrent* -- but what if it's not so clear-cut? It turns out DINOv2 activations can be fit approx. 1-to-1 with only 2 transformer blocks applied recurrently. Leb by @mozesjacobs @thomas_fel_ & @RichieHakim ⬇️

Check out the demos & project page here: xiu-cs.github.io/FMPose3D/

🚨 ✨ New 3D pose estimation method from @mwmathislab! #FMPose3D allows for monocular (i.e. single camera) 2D ➡️ 3D 🔥 Led by @TiwangCS & w/@xiaoyu912huang #FMPose3D is SOTA on human & animal 3D benchmarks, & will be integrated into @DeepLabCut ⬇️👀 📝 arxiv.org/abs/2602.05755

Neural interfaces are the winners of the 2026 Queen Elisabeth Prize for Engineering. Nine scientists are honoured, including J.Bloch and G.Courtine (EPF/UNIL/CHUV,) recognized for their development of electronic spinal stimulation technology. actu.epfl.ch/news/neuroscie…

🐻 reID of unmarked Grizzly's in Alaskan wilderness from single images is a daunting task, but in our new work out in @CurrentBiology that is exactly what Beth Rosenberg & Mu Zhou & @TrackingPlumes! 🔥 Check out the work - many years in the making... sciencedirect.com/science/articl…

Check out the news article here: news.epfl.ch/news/ai-enable…

Biologically, it reveals that bears have a much larger range than previously documented. This has important implications for conservation efforts in Alaska ...

🔥🍾 massive congratulations to @shaokaiyeah and Haozhe Qi for #LLaVAction to be accepted to #ICLR2026! TL;DR the first MMLLM for action understanding, + a new benchmark, + methods to train your MMLLMs for such tasks 🕺🏻💃🏼 💻 llavaction.epfl.ch 💙 arxiv.org/abs/2503.18712

How to get AI to make discoveries on open scientific problems? Most methods just improve the prompt with more attempts. But the AI itself doesn't improve. With test-time training, AI can continue to learn on the problem it’s trying to solve: test-time-training.github.io/discover.pdf

@AntonioLozanoDL @NatureNeuro ;) mackenziemathislab.org/publications

@TrackingActions @NatureNeuro Congrats!! Can’t access but would love a copy

👀 How can we use new neuroscience insights to build *adaptive* AI agents and leverage the many foundation models ( which are much like different brain areas)? Check out my pitch in @NatureNeuro in the Jan 2026 issue 💚 ⬇️ nature.com/articles/s4159… #AIagents #neuroscience #AGI

@FlyBottleEscape @Nature @PennMedicine @DeepLabCut 💖💖💖

@TrackingActions @Nature @PennMedicine @DeepLabCut 🙌🏼😎🙏🏼🤠🤘🏽 It all started w DLC !!! So THANN YOU!

🧠⚡️💊New @Nature publication ! Mimicking opioid analgesia in cortical pain circuits We built a brain-behavior framework to decode spontaneous chronic pain in mice—and to biologically mimic morphine with a synthetic opioid gene therapy nature.com/articles/s4158… @PennMedicine