Life is too short to worry about little things. Have fun. Fall in love. Regret nothing, and don't let people bring you down. Study, think, create, and grow. Teach yourself and teach others.

Five years ago, thanks to the leadership of Matthew Lungren MD MPH, @stanfordAIMI released the CheXpert images: 223K JPG CXRs with labels for 14 conditions. CheXpert has been cited >6000 times, mostly related to development of supervised learning methods. Much has changed since then.🧵

too many preference optimization methods in the last two months. need to optimize my preferences for learning which ones in depth.

Checkout this CT VLM work where we train ~15k CTs using paired diagnostic codes from EHRs, and radiology reports. The model shows strong off-the-shelf adaptability to several downstream tasks! Work led by Louis Blankemeier 😀!

It amazes me that so many (even tech) people still don't get the transformative power of generative AI. The best text to image and video models in existence today all use synthetically generated captions (see e.g. Dalle 3 paper). Human generated data is substantially inferior.

Disruptive Autoencoders was shortlisted for the best oral paper at MIDL2024!! Congratulations Jose! Jeya Maria Jose JHU ECE Paper: openreview.net/pdf?id=8vtxQox… Code: github.com/Project-MONAI/… Weights: github.com/Project-MONAI/… Photo credit: Amar Kumar

Career update: Happy to share I recently joined Microsoft Research as a senior researcher 🙂.

What is intelligence? The ability to know there is never going to be a right answer to this question.

RebotNet was accepted at WACV 25 WACV , architecture code here: github.com/jeya-maria-jos… Thanks AK for sharing our work!

ML twitter this morning 😂

Microsoft Research’s AI2BMD, an AI-based system that efficiently simulates a wide range of proteins in all-atom resolution, can advance drug discovery and biomolecular research. msft.it/6015WkLtR

Interested in research on multi-modal AI and its applications in health? Consider applying for internship at MSR - Health Futures 👇

I wish overleaf was smart enough to automatically cite relevant papers as I write.

🎉 Excited to share that our latest research, 𝘛𝘪𝘮𝘦-𝘵𝘰-𝘌𝘷𝘦𝘯𝘵 𝘗𝘳𝘦𝘵𝘳𝘢𝘪𝘯𝘪𝘯𝘨 𝘧𝘰𝘳 3𝘋 𝘔𝘦𝘥𝘪𝘤𝘢𝘭 𝘐𝘮𝘢𝘨𝘪𝘯𝘨, has been accepted at 𝗜𝗖𝗟𝗥 2025! 🚀 🔍 𝗜𝗺𝗽𝗿𝗼𝘃𝗶𝗻𝗴 𝗠𝗲𝗱𝗶𝗰𝗮𝗹 𝗜𝗺𝗮𝗴𝗲 𝗣𝗿𝗲𝘁𝗿𝗮𝗶𝗻𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗧𝗶𝗺𝗲-𝘁𝗼-𝗘𝘃𝗲𝗻𝘁

Waiting for superhuman’s first exam.

it’s incredible to think that we’re living in an era with more opportunities than ever before, where so many people have the tools and resources to do what they love.

💥 New paper: Think Before You Diffuse Meet DiffPhy — LLM-guided, physics-aware video diffusion 🎥🧠🌍 SOTA on real-world motion & dynamics! 🔗 bwgzk-keke.github.io/DiffPhy/ JHU Computer Science Johns Hopkins Data Science and AI Institute Johns Hopkins Engineering Yiqun Mei #DiffusionModels #VideoGeneration