If you'd like to work with me, @PascalMettes and @cgmsnoek on representational alignment of video-AI with human cognition and neuroscience 📽️🧠🤖🧑‍💻, you should apply for this specific position 👇 vacatures.uva.nl/UvA/job/PhD-Po… We look forward to your application!

Applications now open- 2nd Summer of Research @BristolUni #MachineLearning and #ComputerVision (MaVi) group. R u PhD student with overlapping interests to us? U can visit for 3mnths this summer! Apply - DL 19/1/24 uob-mavi.github.io/[email protected]… Watch 2023 cohort: youtu.be/p12V1z2Te0Y

📢 Applications for OxML 2024 are now open! 🌟 Apply now: oxfordml.school Join the ranks of ML enthusiasts diving into two cutting-edge courses: MLx Representation Learning & Generative AI: Delve into the complexities of neural networks, computer vision, and more. MLx… twitter.com/i/web/status/1…

Are you interested in performing high-impact interdisciplinary research in Artificial Intelligence and its alignment with humans and society in the context of public safety? Come work with us @cgmsnoek & an excellent interdisciplinary team in Amsterdam: vacatures.uva.nl/UvA/job/PhD-Po…

Our hyperbolic learning software library has been accepted to #ACMMM2023 and will compete in the Open-Source Software Competition tomorrow! Check out the code, with API and tutorials, here: …bolic-learning-library.readthedocs.io Paper: arxiv.org/abs/2306.06154 twitter.com/MvanSpengler/s…

🚀 One of the highlights of this #UvADataScienceDay is the launch of the DSC HAVA-Lab by PI @cgmsnoek. A first of its kind research laboratory uniting forces from all 7 UvA faculties for an ethical approach to #videoAI 🎥

Excited to share that our previously announced Hyperbolic Learning Library and the corresponding paper "HypLL: The Hyperbolic Learning Library" have been accepted for the open-source software competition at #ACMMM2023! w/ @phippli and @PascalMettes 1/3

@fernandopsantos @ERC_Research @UvA_Amsterdam Great news Fernando!

Happy to share that our paper "Poincaré ResNet" has been accepted for #ICCV2023! w/ Erwin Berkhout and @PascalMettes 1/5

@geleonti Thanks for the kind words!

Our work on residual networks in hyperbolic space from the first layer is accepted to #ICCV2023! Hyperbolic learning will transform how we perform hierarchical and robust training and inference, this paper already shows great potential over standard learning. See thread below: twitter.com/MvanSpengler/s…

I had fun visiting Hawai’i for #ICML 2023 and presenting our work on  “Hyperbolic Image-Text Representations”! 🌋🏖️Code and pre-trained models are now released! Paper: arxiv.org/abs/2304.09172 Code: github.com/facebookresear… @metaai @michigan_AI

The MIT Climate and Sustainability Consortium (impactclimate.mit.edu) under its Impact Fellows program (impactclimate.mit.edu/get-involved/i…) is hiring a postdoctoral fellow with a technical focus on computing and climate!

One week to the registration deadline, and two weeks to the paper submission deadline. Please consider joining the NCCV 2023 (The Netherlands Conference on Computer Vision) in the Netherlands. sites.google.com/view/nccv2023/…

I'm thrilled to have been awarded a Veni grant! My project aims to recognize how actions happen by perceiving subtle details in videos. twitter.com/NWOFunding/sta…

Our solution is to take a transductive view by matching the distributions of embedded actions and videos. We do so through a hyperspherical optimal transport. The proposed approach is generic and works for any zero-shot method, improving recognition and localization. 3/3

Look at the figure below. The orange line denotes how often each action is selected during testing for a SOTA large video model. This is a clear long-tailed distribution with a few popular actions. As such, the accuracy for most actions is low or zero, hampering accuracy. 2/3

Large video models enable top zero-shot action recognition scores, but they are biased towards a few unseen action. In some cases, a quarter of the actions are never selected in the first place (!) In our new #IJCV paper, we provide a solution: rdcu.be/dhgVV 1/3

📣Please RT!📣 I have a PhD opening @UvA_IvI @UvA_Amsterdam as part of the #ELLISunitAmsterdam! Looking for an enthusiastic MSc with an #ML background interested in studying the interface of #cogneuro and #AI🧠↔️🤖 Apply before Aug. 26th! More info👇vacatures.uva.nl/UvA/job/ELLIS-…

I am extremely excited to announce an opening (Ph.D. candidate) in my group at @TUeindhoven: generativeai-tue.github.io Please apply here: jobs.tue.nl/en/vacancy/phd… Join us in the fantastic journey of developing #GenerativeAI!

@AbirDasUCR @cgmsnoek @UvA_Amsterdam Hi Abir, sorry it was in-person only.

We want pretrained models that output embeddings *and* uncertainties. So, we've designed a new benchmark: Uncertainty-aware representation learning (URL). 📄 arxiv.org/abs/2307.03810 💻 github.com/mkirchhof/url Four findings below. 🧵1/6

Exciting new #lab at @uva_dsc on human-aligned #video #AI, Vacancies for 7 #PhD students expected to open in the fall. We are looking for a special breed of multidisciplinary talent, who combine AI-tech with AI-society expertise. #Amsterdam #AI twitter.com/uva_dsc/status…

📢Exciting news! This new DSC #research programme will establish a unique laboratory that studies video #AI from a multidisciplinary perspective – the research investigation team will consist of investigators and PhD students from all 7 UvA faculties 👇 dsc.uva.nl/content/news/2…

I am super happy that our work <Query by Activity Video in the Wild> has been accepted by ICIP this year, joint work with @twuilliam @PascalMettes and @cgmsnoek more details will follow~😊😊😊

Thrilled to share our ICMR 2023 paper to deal with the long-tailed and imbalanced predicate distribution in scene graph generation in videos. Paper: arxiv.org/abs/2306.10122. Code: github.com/shanshuo/ML-MWN. @yingjundu2 @PascalMettes @cgmsnoek

@zhengyiluo Hi Zen, thanks and apologies. There were mixed instructions unfortunately, so there was no audio for the virtual part. We plan to make all the slides, as well as the code tutorial available on our website soon, so that all important information is there.

📣📣 Hear, hear @doughty_hazel is hiring for her new video understanding team at @UniLeiden, highest recommendation! #CVPR2023 twitter.com/CSProfKGD/stat…

Several team members of the Video & Image Sense Lab are at #CVPR2023 in #Vancouver this week. Meet and greet us to discuss our latest computer vision by learning research. Three papers in the main program on #foundation #models, #generative #ai and #long-…lnkd.in/eTKCZjbE

Tomorrow at 9:00 PT, we start our #CVPR2023 tutorial on hyperbolic learning. We hope to see you there. We'll cover the state of the field, mathematical basis, hierarchical inspirations, and a code tutorial. All the tools you need to get started! Website: sites.google.com/view/hdlcv-cvp…