Sören Becker retweeted
Dealing with undesired distribution shifts in unpaired translation tasks? Our #ICLR2024 paper shows how to mitigate them leveraging Unbalanced OT!
We propose a method to incorporate unbalancedness into any neural Monge map estimator and demonstrate the benefits of unbalancedness.
Sören Becker retweeted
Existing datasets can be combined and re-used to bring insights unattainable from individual datasets. To see how to achieve this, even in the presence of substantial batch effects, join my talk at M2D2 series hosted by @_portal_ on 2024/02/13 at 5pCET. shorturl.at/cxJS8
Sören Becker retweeted
All are invited to submit to our first ICLR 2024 workshop on AI4DifferentialEquations in Science! We welcome submissions that push forward the use of AI for solving ODEs and PDEs with applications in earth sciences, weather, climate and beyond.
#ai4science #iclr2024
Sören Becker retweeted
Join us for a discussion on uncovering new biological insights through the application of Optimal Transport. We'll start with discrete OT and continue with novel Neural OT algorithms using Flow Matching! Can't wait? Feel free to check out moscot-tools.org!
#DynamicsClub in January: 𝗗𝗼𝗺𝗶𝗻𝗶𝗸 𝗞𝗹𝗲𝗶𝗻 @Dominik1Klein @HelmholtzMunich will share his work with @fabian_theis on "Optimal transport in single-cell genomics with moscot".
Jan 10 (Wed) at 11am PT / 2pm ET / 8pm CET. Join us on Zoom: bit.ly/3X9LUgj
Sören Becker retweeted
We are very excited to present the development of Zman-seq (“Zman”, Hebrew for “time”), the 1st technology that measures single-cell transcriptomes and physical time in vivo, led by @D_Birschenkaum, @CuriousKX, @FlorianIngelfi1, @AssafWeiner
sciencedirect.com/science/ar…. (1/19)
Sören Becker retweeted
Is binarization of scATAC-seq data really necessary? The conclusion from our analysis is that a quantitative treatment is in fact beneficial. Now out in Nature Methods! @gagneurlab @fabian_theis nature.com/articles/s41592-0…
Many additions since the preprint 👇(1/n)
Sören Becker retweeted
The AI Transparency Institute is hosting the AI governance forum virtually on 1 December. The speakers come from diverse backgrounds, including law, policy, computer science and philosophy! Do drop by!
Register: lnkd.in/dKV-zbuV
Speakers: @xriskology @sayashk @ZennerBXL
Really enjoyed this collaboration!
🚨 ODEFormer is on Arxiv! arxiv.org/pdf/2310.05573.pdf
We show that Transformers can recover the differential equations governing dynamical systems from noisy & irregularly sampled trajectories.
Very fun collaboration with @SorenBecker, @TrackingPlumes, @pschwllr & @k__niki!
🧵⤵️
Sören Becker retweeted
Have you ever wondered where do #causal variables come from? 🤔
Join the discussion on #causal #representation #learning at our #CRL workshop @NeurIPSConf
Submission deadline: Sept 29
Workshop date: Dec 15/16
🚨CFP #Causal #Representation #Learning @NeurIPSConf 2023 WS🚨
Do you have ideas on how to learn #causal representations from raw data, e.g. images, #abstract #macro variables or leverage #causality in #ML?
Then submit a max 6 pg paper to our workshop crl-workshop.github.io/
Sören Becker retweeted
1/n We're very excited to announce the release of #DeepRVAT, a deep neural network approach to learn burden scores from rare variants by integrating dozens of annotations in a data-driven manner.
Sören Becker retweeted
1/10 Looking for a tool to map cells across time and space? We introduce moscot-tools.org, a scalable framework for #optimaltransport (#OT) applications in single-cell genomics! bit.ly/3MiICFc
Sören Becker retweeted
Glad to see our review on best practices for single-cell analysis across modalities out @NatureRevGenet! In a big team effort led by @LukasHeumos & @AnnaCSchaar, we recommend workflows based on benchmarks. Paper at rdcu.be/c8RU5 & extension at sc-best-practices.org.
Sören Becker retweeted
The Munich Center for Machine Learning (MCML) invites applications for multiple roles:
PhD positions: mcml.ai/opportunities/fundin… (let them know when you want to work with me)
Group Leader ML: mcml.ai/opportunities/fundin…
Group Leader Ethics of AI/ML: mcml.ai/opportunities/fundin…
Sören Becker retweeted
Very excited to be presenting our work later today at the @lmrl_bio workshop @NeurIPSConf as a contributed talk! For further discussion, join me at poster 43 during the first poster session.
Interested in single-cell trajectory inference? Or Parameterized Monge Maps? Check out our paper "Modeling Single-Cell Dynamics using Unbalanced Parameterized Monge Maps", now accepted at @lmrl_bio ! biorxiv.org/content/10.1101/…
Sören Becker retweeted
We would like to thank everyone that followed our workshop, in person or remotely! It was a pleasure to meet all of you 🤗
Thanks to all the speakers for their insightful talks. Excited about further developments on the interaction between causality and dynamical systems! 🔄
Sören Becker retweeted
Delivered a presentation on Markov properties for continuous-time dynamical systems at #NeuRIPS22 workshop "Causal View on Dynamical Systems". In case you missed it, you can watch it here: recorder-v3.slideslive.com/?…
Sören Becker retweeted
I'm very excited to finally announce our online book sc-best-practices.org on single-cell analysis best practices. We build upon the excellent work of @MDLuecken and expand his tutorial to more analysis use-cases and all widely used modalities.
1/n Single-cell data analysis is complex with many tools to choose from - but which ones work best and should be used when? Led by @lukasheumos and @AnnaCSchaar, we wrote a free online book that aims to guide single-cell data analysts: sc-best-practices.org
Sören Becker retweeted
You are a PhD student and need financial support 💰 to pay the conference + workshop fees to attend the CDS workshop? Please emails us ([email protected]) - we have some funding available and might be able to help. Please add your CV and a statement of motivation.
Sören Becker retweeted
🤔 Wondering how to use uncertainty quantification for biological insight? 🧬💡
Checkout our arXiv pre-print on:
"Uncertainty Quantification for Atlas-Level Cell Type Transfer"
arxiv.org/abs/2211.03793
Let's dig in! 🧵👇 1/
