Transfer learning is a widespread technique used in deep learning. It consists in using knowledge learned from one source task/domain in another target task/domain. It has been studied in-depth in computer vision, natural language, and speech, but what about brain data?

I've had a great time leading the ML4H Author Mentorship Program which ends on Sept 7th. 📢For those who didn't have a chance to participate, the Reviewer Mentorship Program kicks off soon and is still looking for mentors. ⏳Sign up by Aug 30!

🚨 New dataset! I'm releasing a synthetic chest X-ray dataset on Hugging Face datasets. This was created as part of a paper entitled Privacy Distillation (more details soon). We ensured the images can NOT be used for re-identification! huggingface.co/datasets/SinKo…

📣Exciting news! Registration for #ML4H2023 is now OPEN! Be part of the action on Dec 10 at the Intercontinental New Orleans. Hurry, online registration closes on Dec 3! Secure your spot now: ahli.cc/ml4h2023/

⏳ Only 3 days left to join our Career Mentorship Program! Want insights from experienced researchers or eager to share your wisdom with budding students? Sign up now! 🔗 Mentees: forms.gle/xpAzczgHvAfk3d… 🔗 Mentors: forms.gle/VZAU9G4Lq8urBs… Hurry, the deadline is Oct 15, 2023!

📢 New dataset on Hugging Face! 🔬 Synthetic mammography (medical imaging) dataset. These patients do not exist. 🎯 100k synthetic mammograms. 512x512 size. 3 labels: - Low masking level (score ≤ 2) - Medium masking level (2 < score ≤ 6) - High masking level (score > 6) 🩺

I am looking forward to sharing our findings at the Deep Generative Models for Health workshop at NeurIPS 2023! It was such a pleasure to work with Bruno Aristimunha on this project! 🧠👨‍💻 DGM4H Workshop @ NeurIPS 2023 #NeurIPS2023 #EEG #ProudMoment

#PhD in #AI for development of digital twins of human brain functional dynamics for diagnosis and treatment of neuropsychiatric disorders with Robert Leech as part of Kings Digital Twins for Health CDT. Pls RT lnkd.in/eApKT_fK

Attention neuroscientsts! 🧠 We just launched Neurosynth Compose: A free and open platform for neuroimaging meta-analysis. NS-Compose makes it easy to perform custom neuroimaging meta-analyses without leaving the browser. It's live, check it out! compose.neurosynth.org

How predictable is neuroscience? Can LLMs outperform humans? Please participate in the BrainGPT.org survey to help us find out. You choose between two versions of a neuro abstract: the original vs. one with altered results. Which is which? research.sc/participant/lo… 1/2

Thank you to the ML4H 2023 Outreach Committee! Harry Reyes Nieva Ghada Zamzmi Jessica Dafflon🍍 Caleb Ellington Sarah Jabbour Hyewon Jeong Yuzhe Yang

"Large language models surpass human experts in predicting neuroscience results" w Xiaoliang (Ken) Luo and BrainGPT.org. LLMs integrate a noisy yet interrelated scientific literature to forecast outcomes. arxiv.org/abs/2403.03230 1/6

Are you planning to submit a paper to ML4H 2024 and would like to receive mentorship from senior scientists? Are you interested in mentoring early-stage researchers? Sign up for the ML4H Submission Mentorship Program by July 5th AoE! Program details: ahli.cc/ml4h/mentorshi…

📢Exciting job opportunity! 🧠 My lab is searching for a Research Assistant to work with computational modelling and neuroimaging to study the mechanisms of compulsivity 🔗Apply here: cis7.bbk.ac.uk/vacancy/resear… 🗓️ Closing date 16th July 🔁 Please retweet / share

Reminder to sign up to be a mentor or a mentee- deadline is this Friday, July 5th!!

Sign up for the Reviewer Mentorship Program at #ML4H2024! 🧐 Want to review an ML4H paper but don't feel confident yet to do it alone? OR 👩‍🏫 Passionate about guiding early-stage researchers in the review process? Sign up by August 30 AoE! 📝 Apply here: forms.gle/79BU6VPXJpqEDK…

Calling all potential mentors! Many talented early-career students have signed up for our mentorship programs, and they need you! For both reviewer mentorship & career mentorship! Sign up before Aug 30!

In biology, unpaired data is the norm as experiments are often destructive. This limits our ability to build powerful multimodal models for drug discovery. To fuel a multimodal future in TechBio, we propose a simple algorithm for matching. 🧵 youtu.be/7a29mGz9LXI

Introducing a small portion of proprietary phenomics data from Recursion into the training of MolGPS, our foundation model for molecular property prediction, led to massive performance increases on new & existing benchmarks. Learn more about the impact of data & scale.

"Large language models surpass human experts in predicting neuroscience results" w Xiaoliang (Ken) Luo and BrainGPT.org. LLMs integrate a noisy yet interrelated scientific literature to forecast outcomes. nature.com/articles/s4156… 1/8