Lightnews — Scholar-powered news

Reposted by Jakob Macke

C3N @c3neuro.bsky.social · 8d

Hello Frankfurt 🇩🇪 - We are excited to share the latest result from our group and collaborators at
@bernsteinneuro.bsky.social 🚀🧠 #BernsteinConference

Thanks to @brainloops.bsky.social, @ekfstiftung.bsky.social and @cherish-msca.bsky.social for supporting these projects and our scholars.

1 6 12

Reposted by Jakob Macke

Machine Learning in Science @mackelab.bsky.social · 8d

The Macke lab is well-represented at the @bernsteinneuro.bsky.social conference in Frankfurt this year! We have lots of exciting new work to present with 7 posters (details👇) 1/9

1 9 29

Reposted by Jakob Macke

sbi - Simulation-based inference @sbi-devs.bsky.social · 29d

From hackathon to release: sbi v0.25 is here! 🎉

What happens when dozens of SBI researchers and practitioners collaborate for a week? New inference methods, new documentation, lots of new embedding networks, a bridge to pyro and a bridge between flow matching and score-based methods 🤯

1/7 🧵

1 16 28

Reposted by Jakob Macke

Seth Axen 🪓 @sethaxen.com · Jul 31

Sharing this here a bit late, but @vstaros.bsky.social and I wrote a little something about our experience contributing to the @sbi-devs.bsky.social (simulation-based inference) hackathon. @mlcolab.org @mackelab.bsky.social

We were obviously very hungry while writing.

A retrospective on the 2025 SBI Hackathon

You walk into a bakery, take one bite of a still-warm pastry, and think: “Whoa - there’s rye flour, a hint of orange zest, maybe cardamom… and is that buckwheat honey?” From that single taste you begi...

mlcolab.org

3 11

Reposted by Jakob Macke

Daniel Gedon @danielged.bsky.social · Jul 23

My first paper on simulation-based inference (SBI) as part of @mackelab.bsky.social!

Exciting work on adapting state-of-the-art foundation models for posterior estimation. Almost plug-and-play, and surprisingly effective.

Paper/code in thread below 🧵

Machine Learning in Science @mackelab.bsky.social · Jul 23

New preprint: SBI with foundation models!
Tired of training or tuning your inference network, or waiting for your simulations to finish? Our method NPE-PF can help: It provides training-free simulation-based inference, achieving competitive performance with orders of magnitude fewer simulations! ⚡️

2 18

Jakob Macke @jakhmack.bsky.social · Jul 23

I have been genuinely amazed how well tabpfn works as a density estimator, and how helpful this is for SBI ... Great work by @vetterj.bsky.social, Manuel and @danielged.bsky.social!!

Machine Learning in Science @mackelab.bsky.social · Jul 23

New preprint: SBI with foundation models!
Tired of training or tuning your inference network, or waiting for your simulations to finish? Our method NPE-PF can help: It provides training-free simulation-based inference, achieving competitive performance with orders of magnitude fewer simulations! ⚡️

3 16

Reposted by Jakob Macke

ML for Science @ml4science.bsky.social · Jul 16

Together with #AIMS, the African Institute for Mathematical Sciences, we have an exciting position to fill: The AIMS - Tübingen Junior Research Chair in Machine Learning for Science! 1/2

Formulas and numbers from the field of machine learning written in white chalk on a blackboard

1 8 9

Reposted by Jakob Macke

ML for Science @ml4science.bsky.social · Jul 16

Come work with us in Tübingen! Our new professor @mariokrenn.bsky.social is looking for PhDs and postdocs! The group builds #AI systems for discovering new concepts, experiments and ideas in #physics. Find out more: uni-tuebingen.de/en/128980#c2... #AIforScience

Mario Krenn (right) and eight members of his group stand next to each other for a group photo, with green plants in the background.

3 12

Reposted by Jakob Macke

Anna Beyeler 🐝 @anna-beyeler.bsky.social · Jul 10

👨‍💻 Open PI position in our institute 👨‍💻 !! If you are an expert in Computational Neuroscience and want to start your lab in Bordeaux, contact us !
www.fens.org/careers/job-...
@neuromagendie.bsky.social
@neurobordeaux.bsky.social

45 49

Jakob Macke @jakhmack.bsky.social · Jun 12

My point is not that you should include our work — but that more computational folks should actually tackle this wonderful data now that it is abailable, with ML or without (although we certainly have found ML useful)!

1

Jakob Macke @jakhmack.bsky.social · Jun 11

Sounds interesting! Loved how your pitch includes the availability of the drosophila connectome, unfortunately none of the abstracts seem to mention it?

1

Jakob Macke @jakhmack.bsky.social · Jun 11

Sbi can take you places … very proud about @gmoss13.bsky.social et als work, a wild and serendipitous collaboration made possible by @ml4science.bsky.social!

Guy Moss @gmoss13.bsky.social · Jun 11

Have I been to Antarctica? No. But my colleagues have, and we can learn a lot from the data they collected! Really happy to share that our work is now published!

Machine Learning in Science @mackelab.bsky.social · Jun 11

Thrilled to share that our paper on using simulation-based inference for inferring ice accumulation and melting rates for Antarctic ice shelves is now published in Journal of Glaciology!

www.cambridge.org/core/journal...

6

Jakob Macke @jakhmack.bsky.social · Jun 6

Oh I am more optimistic and even even excited, one can disect it, and it has interpretable units and sub circuits … we might need better tools though to make it easier for people to play with it!!

3

Jakob Macke @jakhmack.bsky.social · Jun 6

was our paper used as evidence for or against this thesis? (My biased view is that this approach shows how one can big data and mechanistic insights can be friends …)

1 4

Reposted by Jakob Macke

François Rozet @francois-rozet.bsky.social · Jun 3

I always find the figures from the @mackelab.bsky.social really clear and pretty!

For example in arxiv.org/abs/2404.09636.

All-in-one simulation-based inference

Amortized Bayesian inference trains neural networks to solve stochastic inference problems using model simulations, thereby making it possible to rapidly perform Bayesian inference for any newly obser...

arxiv.org

1 1 13

Reposted by Jakob Macke

Basile Confavreux @bconfavreux.bsky.social · Jun 4

We’ve successfully automated part of the (neuro)scientific process. Now I may be out of a job. After a lot of iterations, here’s our framework for automatic discovery, built to embrace degeneracy and the realities of underconstrained modeling.

Tim Vogels @tpvogels.bsky.social · Jun 2

We just pushed “Memory by a 1000 rules” onto bioRxiv, where we use clever #ML to find #plasticity quadruplets (EE, EI, IE, II) that learn basic stability in spiking nets. Why is it cool? We find 1000s!! of solutions, and they don’t just stabilise. They #memorise! www.biorxiv.org/content/10.1...

Memory by a thousand rules: Automated discovery of functional multi-type plasticity rules reveals variety & degeneracy at the heart of learning

Synaptic plasticity is the basis of learning and memory, but the link between synaptic changes and neural function remains elusive. Here, we used automated search algorithms to obtain thousands of str...

www.biorxiv.org

2 6 32

Jakob Macke @jakhmack.bsky.social · Jun 3

Thanks! Trying to make good figures is an important part of the @mackelab.bsky.social experience ;-). Indeed, this design was initiated by @janmatthis.bsky.social and then passed on and refined by @deismic.bsky.social and others! We still use it, although sometimes in white-on-black for talks ...

5

Jakob Macke @jakhmack.bsky.social · Jun 2

magic, by magician @bconfavreux.bsky.social, using a magic SBI-wand conjured with Poornima Ramesh, under the auspices of the wise @tpvogels.bsky.social!

Tim Vogels @tpvogels.bsky.social · Jun 2

We just pushed “Memory by a 1000 rules” onto bioRxiv, where we use clever #ML to find #plasticity quadruplets (EE, EI, IE, II) that learn basic stability in spiking nets. Why is it cool? We find 1000s!! of solutions, and they don’t just stabilise. They #memorise! www.biorxiv.org/content/10.1...

Memory by a thousand rules: Automated discovery of functional multi-type plasticity rules reveals variety & degeneracy at the heart of learning

Synaptic plasticity is the basis of learning and memory, but the link between synaptic changes and neural function remains elusive. Here, we used automated search algorithms to obtain thousands of str...

www.biorxiv.org

13

Reposted by Jakob Macke

Nima Dehghani @neurovium.bsky.social · May 8

Last month, we organized #ODIN2025 @alleninstitute.org (along w Karel, Saskia @sejdevries.bsky.social , David & Satra) with the theme "Integrating Scales & Modalities".
See the videos📣👇
Full playlist: youtube.com/playlist?lis... 🎥
Speaker: alleninstitute.org/odin-symposi...
youtu.be/N1IMki0RnIc?...

Opening remarks Karel Svoboda & Nima Dehghani

YouTube video by Allen Institute

youtu.be

3 9

Reposted by Jakob Macke

TübingenNeuroCampus @tueneurocampus.bsky.social · May 12

📢 Amazing speaker in our Neurocolloquium on Wed. May 21 at 4.15pm CEST: Marion Silies @unimainz.bsky.social‬ will talk about "From heterogeneous wiring to degenerative function in motion-detection circuit", live in Tübingen & in Zoom.
🙌 Host: @jakhmack.bsky.social
👉 Sign up: tinyurl.com/32ku5jth

2 6

Jakob Macke @jakhmack.bsky.social · May 3

Nice article in the transmitter about connectome-enabled brain models, covering @lappalainenjk.bsky.social’s work with @srinituraga.bsky.social and colleagues!!

The Transmitter @thetransmitter.bsky.social · May 2

It took more than a decade to trace the 130,000 connections in a fruit fly's brain. With the map in hand, researchers are turning to the next step in connectomics: building simulations.

By @ldattaro.bsky.social

#neuroskyence

www.thetransmitter.org/connectome/c...

Connectomics 2.0: Simulating the brain

With a complete fly connectome in hand, researchers are taking the next step to model how brain circuits fuel function.

www.thetransmitter.org

1 5 9

Reposted by Jakob Macke

Jan-Matthis Lueckmann @janmatthis.bsky.social · Apr 24

We'll present our #ICLR2025 spotlight on ZAPBench this afternoon: 📍 Hall 3 #61!

7 22

Jakob Macke @jakhmack.bsky.social · Apr 25

Most simulators for time-series data (e.g., numerical solvers for differential equations) are Markovian-- this can be exploited for efficient simulation-based inference on time-series data! Talk to Manuel and Shoji at #ICLR2025, or read the paper ⬇️!

Machine Learning in Science @mackelab.bsky.social · Apr 25

Excited to present our work on compositional SBI for time series at #ICLR2025 tomorrow!

If you're interested in simulation-based inference for time series, come chat with Manuel Gloeckler or Shoji Toyota

at Poster #420, Saturday 10:00–12:00 in Hall 3.

📰: arxiv.org/abs/2411.02728

Compositional simulation-based inference for time series

Amortized simulation-based inference (SBI) methods train neural networks on simulated data to perform Bayesian inference. While this strategy avoids the need for tractable likelihoods, it often requir...

arxiv.org

1 1 19

Reposted by Jakob Macke

Auguste Schulz @auschulz.bsky.social · Apr 17

Thanks so much for the shout-out, and congrats on your exciting work!! 🎉 🙂

Also, a good reminder to share that our work is now out in Cell Reports 🙏🎊

⬇️

www.cell.com/cell-reports...

1 10 36

Jakob Macke @jakhmack.bsky.social · Mar 28

I am also not at Cosyne ;-). Then send us an email to [email protected], full instructions of what we need are here: www.mackelab.org/jobs/

Jobs - mackelab

The MackeLab is a research group at the Excellence Cluster Machine Learning at Tübingen University!

www.mackelab.org

1 1