Claudio Michaelis

building polybot, a flexible robot for a sustainable agriculture future

[Claudio Michaelis @clmich]

Blows my mind that our POLYBOT made it into the opening of #COP27 🤯

[Wieland Brendel @wielandbr]

Yeah! Our work on an intelligent #robot for #sustainable #farming (POLYBOT) made it into the opening of #COP27 World Leader Summit in the talk of @VeronikaEyring. Interested in bold climate tech solutions? Talk to me or @clmich at #COP27 or DM us.@Cyber_Valley @MPI_IS @ml4science

[Wieland Brendel @wielandbr]

Are you into cutting-edge #HPC/#AI clusters? The Tübingen AI Center, one of Europe's top #ML research institutions, searches for a senior expert to lead its #AI cluster (> 1000 GPUs, > 200 users) team and design an upcoming large-scale upgrade:
https://t.co/tCKfPA5UfD

[Shift Happens Workshop ICML 2022 @shifthappens_22]

Are you interested in the generalization capabilities of ImageNet classifiers?

Then attend our workshop! Join us at the ShiftHappens Workshop @icmlconf #ICML2022 this Friday (22 July) 9am - 7:15 pm EST in Ballroom 4!🎉
https://t.co/NIC1obZVUo 🧵[1/3]

[Claudio Michaelis @clmich]

Come work with us on #robotics for #sustainable agriculture! 🤖🌱

[Wieland Brendel @wielandbr]

Just started a group at Max Planck Institute for Intelligent Systems (@MPI_IS) & Tübingen AI Center🎉 Focus: robustness, identifiable representation learning & interpretability. Plus a cool project on #AI, #robotics & #sustainable food production. Interested in a #postdoc? DM me!

[Shift Happens Workshop ICML 2022 @shifthappens_22]

Contribute datasets📷, metrics📊 and tasks🦾 to a community-built, open source benchmark for computer vision models: Submission for the workshop extended until May 27! *Register your abstract until May 17.*

🔗Details: https://t.co/xzdhu1ziPS
🚀Submission: https://t.co/yecDzRqMwJ

[Max Planck Institute for Biological Cybernetics @MPICybernetics]

#CaCTüS offers a paid 3-months research #internship in Tübingen open to students who fled Ukraine 🇺🇦 and are now in Germany 🇩🇪 . The project requires some programming experience and interest in data analysis & the brain. More: https://t.co/BmOoFN4mru
#ScienceForUkraine

[Mackenzie Weygandt Mathis, PhD @TrackingActions]

📣Part 1 of our quest to better understand the brain was @DeepLabCut.
🔥🦓Now Part 2: Introducing #CEBRA to jointly model neural dynamics & behavior with self-supervised learning. Hypothesis- or data-driven, highly consistent, decodable neural latents
https://t.co/fEiaTSVO2L
🧵👇

[Shift Happens Workshop ICML 2022 @shifthappens_22]

We are thrilled to announce the ShiftHappens Workshop at #icml2022 @icmlconf in which the community builds a challenging open-source benchmark suite for ImageNet-scale models! All accepted submissions will be part of the open-source benchmark suite. [1/6] https://t.co/NIC1obZVUo

[Claudio Michaelis @clmich]

@giffmana @hendrycks @ylecun @imisra_ @__kolesnikov__ BiT is such an amazing paper! I'm referring people to the appendix all the time. (And few-shot learning people to the paper in general, to argue we should leave miniImageNet behind.)

[Claudio Michaelis @clmich]

@oripress1 Please @oripress1, you know he wants all three of them and still won't be happy.

[Bethge Lab @bethgelab]

In 2015, ResNets reportedly surpassed human-level performance on #ImageNet. However, a large generalisation gap remained. Which of today’s exciting directions will close the gap: Vision transformers? CLIP? Self-supervised learning? Bigger datasets? Adversarial training? [1/N]

[Kai Kupferschmidt @kakape]

I‘m struck that it still hasn’t sunk in how much #b117 may change the course of this pandemic.

The initial shock about it being more transmissible seems to have worn off. But we are barely beginning to see its real-world impact.

A story: https://t.co/WDX5GmRma4
And a thread

[jasmine collins @jazco]

love this intro 😂

[Thomas Wolf @Thom_Wolf]

A bit late to the party but the « Shortcut Learning » paper by R. Geirhos @jh_jacobsen @clmich R. Zemel @wielandbr, @MatthiasBethge & F. Wichmann was a really enjoyable read!

A nice overview connecting NNet failures, OOD robustness, fairness, bias

Here: https://t.co/uW10k6q0Sa

[Zemel Group @zemelgroup]

Check out our new paper Flexible Few-Shot Learning -- the same object can belong to different classes depending on context. We found unsupervised representation is better than supervised. A short version at NeurIPS metalearn workshop today at 10 EST. https://t.co/fHMGMfcdRb

[Bradley Love @ProfData]

Introducing the embedding space you didn't know you needed: Human similarity judgments for the entire ImageNet (50k images) validation set. Perfect for evaluating representations, including unsupervised models. It's already bearing fruit, w @BDRoads https://t.co/Cl8iTdcHqj (1/3)

[Nature Machine Intelligence @NatMachIntell]

A close look at what happens when deep learning fails reveals an effect of ‘shortcut learning’. Recognizing that this may be a common characteristic of learning systems, artificial and biological, may help making deep learning more robust. Read the paper: https://t.co/V9Tir2re9G

[jörn jacobsen @jh_jacobsen]

Pleased to share that our paper "Shortcut learning in deep neural networks" has been published as a @nature Machine Intelligence Perspective:

https://t.co/FeB3Scrpgt

PDF access without paywall: https://t.co/8lnUCRHCks

[Claudio Michaelis @clmich]

We hope that our findings will also benefit other areas of machine learning as they suggest that we can achieve human-like generalization capabilities by focusing on wide datasest with diverse categories.
[7/7]

[Claudio Michaelis @clmich]

More classes is all you need for one-shot object detection.

It turns out we can almost close the gap between the detection of known and novel objects simply by increasing the number of categories.

https://t.co/jpCEGb3sOA @clmich @alxecker @MatthiasBethge
[1/7]

[Claudio Michaelis @clmich]

Results generally look good even in complex or crowded scenes. While some problems like the high false positive rate (right column) remain, we are confident that our insights will make solving these issues easier as well.
[6/7]