Research scientist @GoogleAI: https://xitter.cf/t.co/whFMXcyxq5 https://xitter.cf/t.co/ycPSxtdB43
Zurich, Switzerland
Joined September 2010
- Tweets134
- Following1K
- Followers463
- Likes1.5K
jan-matthis retweeted
In collaboration with HHMI Janelia & Harvard, we introduce ZAPBench, a whole-brain activity dataset and benchmark with single cell resolution of a larval zebrafish, which will enable development and comparison of more accurate brain activity models. →goo.gle/44rEpsn
jan-matthis retweeted
Can we predict future brain activity in a small vertebrate?🤔
We're releasing ZAPBench⚡️(#ICLR2025 spotlight): a benchmark to forecast activity in a whole vertebrate brain🧠 at single-cell resolution!🐟 70k+ neurons, 3 TB of data & extensive baselines. Connectome is coming!
🧵👇
jan-matthis retweeted
Happy to announce a Student Researcher opportunity in our team @GoogleDeepMind in Mountain View, California ☀️
If you’re excited about foundational research on controllability of visual/3D generative models, please consider applying: deepmind.google/about/career…
jan-matthis retweeted
Introducing TensorStore, an open-source C++ and Python library designed for storage and manipulation of n-dimensional data, which can address key engineering challenges in scientific computing through better management and processing of large datasets. goo.gle/3Ly9e30
jan-matthis retweeted
Simulation-based inference (SBI) has accelerated model-building across many disciplines. It’s also useful for studying cognition: We present Mixed Neural Likelihood Estimation, an SBI method tailored to models of decision-making! By @janfiete @janmatthis @_rdgao @jakhmack
1/5
jan-matthis retweeted
You like GANs, SBI, and the 100-yr anniversary of Jay Gatsby staring at a green light across the Long Island bay? Check out “GATSBI:Generative Adversarial Training for SBI” accepted at ICLR 2022! With @pramesh95 @janmatthis @janfiete @alvorithm @dvdgbg @ppjgoncalves @jakhmack 1/6
jan-matthis retweeted
🪂 Now available! Slides + notebooks for our ongoing #SBI2021 workshop - content by @deismic_, @alvorithm, @janfiete, @janmatthis, @dvdgbg, @ppjgoncalves, reviewing and much help by @alpiges and Stefan Wezel.
@KyleCranmer @HeissenBergsUnc @glouppe @bkmi13 @ml4science @mackelab
ALT All talks and notebook practicals of our simulation-based inference workshop of 2021 - you can click the links to get to the content.
jan-matthis retweeted
Want to learn how to get more out of your simulator with machine learning? Apply now to become one of 20 participants of our 3-day, hands-on workshop on simulation-based inference! Sign up here! → events.hifis.net/event/167/ ←⏬
jan-matthis retweeted
Pyro 1.7 is released: PyTorch 1.9, biological sequence models, streaming MCMC for big models, AutoStructured guide, reparametrizer initialization, 3 tutorials, 8 distributions. Thanks Eli Weinstein @vitaliikl @nning_ola @mt_sokol @kamathhrishi @ordabayevy github.com/pyro-ppl/pyro/rel…
jan-matthis retweeted
The new and improved version of our review paper on normalizing flows is out!
Now published in the Journal of Machine Learning Research @JmlrOrg
jmlr.org/papers/v22/19-1028.…
Check out our extensive review paper on normalizing flows!
This paper is the product of years of thinking about flows: it contains everything we know about them, and many new insights.
With @eric_nalisnick, @DeepSpiker, @shakir_za, @balajiln.
arxiv.org/abs/1912.02762
Thread 👇
jan-matthis retweeted
Multiple circuit settings of the🦞STG achieve the same network activity. But which of those⚙️ are energy efficient?
Find out at #Cosyne21 Poster 3-027 (a+b, 2-4pm + 9-11pm CET) by @deismic_ @ppjgoncalves!
youtu.be/YF1aAE7uqVQ
The answer will surprise you.
↗️ promoted
jan-matthis retweeted
Simulation-based inference algorithms are useful to identify parameters of simulators from data. But there are so many approaches - how do they compare?
We made a benchmark!
With @janmatthis @janfiete @dvdgbg @ppjgoncalves
arxiv.org/abs/2101.04653
sbi-benchmark.github.io
1/n
jan-matthis retweeted
Dream of working with scientists and top ML people on exciting research challenges, from climate change to education?
Looking for a creative & curious researcher at the ML ⇌ Science Colaboratory, part of @ml4science at @uni_tue.
Read on uni-tuebingen.de/en/universi…
Please share!
jan-matthis retweeted
Pyro 1.4 is released with a new module for epidemiological compartmental modeling, 4 new tutorials, 10 new flows/transforms, 6 new distributions, 2 new reparameterizers, improvements to MCMC and SMC, a new init method, experimental RandomVariable syntax github.com/pyro-ppl/pyro/rel…
jan-matthis retweeted
Do you use numerical simulations, but are struggling to find parameters which make your model fit data? Simulation-based inference doi.org/10.1073/pnas.1912789… can come to the rescue. We are releasing sbi, a toolbox to explore new #SBI methods, mackelab.org/sbi 1/6
jan-matthis retweeted
New: "Training deep neural density estimators to identify mechanistic models of neural dynamics” biorxiv.org/content/10.1101/… @ppjgoncalves @janmatthis @deismic_ Nonnenmacher @kaandocal Bassetto @chc1987 @Bill_P_ @SaraAnnHaddad @TPVogels @dvdgbg. Our biggest project so far! Thread:
jan-matthis retweeted
The Pyro team is excited to release NumPyro, a JAX backend for the Pyro probabilistic programming language. The modeling language mirrors Pyro, and inference via HMC and NUTS is very fast. Let us know what you think! github.com/pyro-ppl/numpyro
jan-matthis retweeted
APT uses neural networks to learn posteriors from simulated data.
Building on work by @gpapamak @driainmurray @janmatthis @ppjgoncalves, David + Marcel came up with a new approach (image: the original whiteboard...) which is more efficient + flexible + scalable to high-dim data.
