Diffusion models like #DALLE and #StableDiffusion are state of the art for image generation, yet our understanding of them is in its infancy. This thread introduces the basics of how diffusion models work, how we understand them, and why I think this understanding is broken.🧵

Our single-nucleus cross-tissue cell atlas is out in @ScienceMagazine as part of @humancellatlas: https://t.co/ELvr9Z0gLw w/ Eugene Drokhlyansky, @ShankaraKAnand, @EvgenijFiskin, @ayshwaryas, @francoisaguet, @asegre, Orit Rozenblatt-Rosen, @KArdlie, Aviv Regev & many others 1/

https://t.co/GOLGA2KBtp

Our team in Heidelberg is expanding! Two exciting postdoctoral positions in our lab at the interface of AI/ML & biotechnology. Project 1) AI/ML for decoding cellular networks using perturbation genetics & 2) spatial multi-omics integration in GBM. Please RT & DM for questions.

After a hiatus, a new series of blogs posts. Do differential geometry and algebraic topology sound too exotic for ML? In recent works, we show that tools from these fields bring a new perspective on graph neural networks First post in the series: https://t.co/NOmSC3yDjh

David Sabatini has filed a defamation lawsuit attempting to character assassinate someone who he reveals WAS A STUDENT when he first had sex with her and…(drumroll) there are MULTIPLE other accusers. @mit how’s that tenure revocation case coming along? #IStandWithHer

Excited that our PHAGE-ATAC work is now online in @NatureBiotech ! We engineered nanobody-displaying phages for single-cell protein profiling and present a single-cell platform to measure protein, chromatin and mtDNA genotypes. 1/n https://t.co/BXSjtisnSL

We have been setting up a new seminar series "Seminar on Advances in Probabilistic Machine Learning" with early-career researchers as presenters. The first speaker will be Will Tebbutt from @CambridgeMLG this Wednesday. 🗓 https://t.co/JKA5DfZWSJ @FCAI_fi @CSAalto @UnivHelsinkiCS

Steven Weinberg, one of the most preeminent physicists in history, passed away. Here's his four lessons, published in Nature back to 2003: ☑️ No one knows everything, and you don't have to ☑️ Aim for rough water ☑️ Forgive yourself for wasting time ☑️ learn history of science

We are very excited to share our single-nucleus cross-tissue atlas w/ Eugene Drokhlyansky, @francoisaguet, @ShankaraKAnand, @ayshwaryas, @EvgenijFiskin, @asegre, Orit Rozenblatt-Rosen, @KArdlie, Aviv Regev and many others 1/n

Random AI thought of the day: the ML boom has suddenly put continuous maths centre stage in comp sci departments. When I was a student the curriculum was dominated by discrete math - sets, logic, etc. Fascinating to see the pendulum swing the other way…

I was lucky to have the main stage @NAACLHLT 🤩 last week, and gave a talk on Explorations in Generative ML 🌸 Uploaded the video and the text of the talk - see blog link below👇🏾. Thanks to everyone the for the great Q&A after 🙏🏾🎉🥳. #NAACL2021 https://t.co/RtjKi6O6O4

I'm delighted to share publicly "The Principles of Deep Learning Theory," co-written with @ShoYaida, and based on work also with @BorisHanin. It will appear on the @arxiv on Sunday and will be published by @CambridgeUP early next year: https://t.co/6DBLHZmryb 1/

👏👏👏I wish that was the norm in academia, not publishing just to get one more paper out

Boğaziçi Üniversitesi Yaz okulunda harika bir ders açılıyor: @tolga_birdal ın Non-Euclidean Methods for ML dersi. Dinlemek için sabırsızlanıyorum!

Great tutorial on #Bayesian neural nets by Andrew Wilson, https://t.co/AuH0mev0m8 , am usually cynical about Bayesian tutes because of their knowledge gaps ... but he connects many disparate ideas. #MachineLearning @andrewgwils NB. ensembling has been Bayesian since 1989

I need your help with an ambitious project. Please fill the survey below and/or share it with your colleagues. I know it is a big ask, and I thank you for your help in advance. #AcademicTwitter #AcademicChatter https://t.co/xSFc8uD8l5

This new Review discusses different strategies to engineer next-generation CAR-T cells with enhanced specificity and sustained function for the treatment of solid tumours. https://t.co/KYPFxhGDYk

If you have less than 3 hours to spare & want to learn (almost) everything about state-of-the-art explainable ML, this thread is for you! Below, I am sharing info about 4 of our recent tutorials on explainability presented at NeurIPS, AAAI, FAccT, and CHIL conferences. [1/n]

Yet another challenge for the single cell community