📢📢 Beyond Model Adaptation at Test Time: A Survey by @zehao_xiao. TL;DR: we provide a comprehensive and systematic review on test-time adaptation, covering more than 400 recent papers 💯💯💯💯 🤩 #CVPR2025#ICLR2025
https://t.co/wvWwdQ8F06
All vision-language models should have hyperbolic embeddings. Vision and language are incredibly hierarchical in nature!
See below our latest work on hyperbolic vision-language models that exploit visual compositions through entailment:
🚀 Excited to share LynX! 🦁
🔑 A new method in visual grounding using a Dual Mixture of Experts—LynX enables pretrained VLMs to continuously learn grounding while retaining their image-language capabilities.
📄 Check out the full paper: https://t.co/2qY0jtPsnS
📢 Announcing TVBench: Temporal Video-Language Benchmark 📺 We reveal that widely used Video-Language benchmarks, such as MVBench, fall short in testing temporal understanding and propose an alternative TVBench: https://t.co/uIDyYdPpzF
Today, we're introducing TVBench! 📹💬
Video-language evaluation is crucial, but are we doing it right? We find that current benchmarks fall short in testing temporal understanding. 🧵👇
Excited to announce that today I'm starting my new position at @utn_nuremberg as a full Professor 🎉. I thank everyone who has helped me to get to this point, you're all the best! Our lab is called FunAI Lab, where we strive to put the fun into fundamental research. 😎 Let's go!
🇨🇦 Deeeelighted to share that this work got into #neurips2024. Many thanks to my dear friend and co-author @Dafidofff, as well as the rest of the team.
Solving PDEs in continuous space-time with Neural Fields on cool geometries while respecting their inherent symmetries! 💫💫
The Self-Supervised Learning: What is Next? workshop at @eccvconf had a great turnout with excellent talks. Slides of most talks are available at https://t.co/8VT2uoHwOV (soon all 🤞). Thanks to all attendees, speakers, and co-organizers for making it a fantastic event!
Stop by today and discuss our @eccvconf paper (SelEx) with me, @doughty_hazel, and @cgmsnoek! 🎉
We present self-expertise—an alternative to self-supervision for learning from unlabelled data with fine-grained distinctions and unknown categories.
📍 Poster #89
🕥 10:30 AM
🚀 Excited to present SIGMA at @eccvconf ! 🎉 We upgrade VideoMAE with Sinkhorn-Knopp on patch-level embeddings, pushing reconstruction to more semantic features. With @mdorkenw.
Let’s connect at today's poster session at 4:30 PM, poster number 256, or send us a DM.
🚀 Excited to present our work on Self-Expertise at #ECCV2024 in Milan!
Join us at poster #89 on Friday, Oct 4 at 10:30 AM to see how self-expertise outperforms self-supervision in tackling unknown data in open-world settings! 🌍 #SelfSupervision#GeneralizedCategoryDiscovery
We're happy to be at #ECCV2024 this week thanks to our cooperation with @VISLab_UvA . Check out @melikaayoughi work at the Instance-Level Recognition Workshop the Self-Supervised Learning workshop.
🚀 Excited to announce that our paper "SelEx: Self-Expertise in Fine-Grained Generalized Category Discovery" has been accepted to ECCV 2024! 🎉
Special thanks to my incredible coauthor @MrzSalehi and my amazing supervisors @y_m_asano, @doughty_hazel, and @cgmsnoek🙏.
We are hiring a postdoc! Come work with us in the booming AI ecosystem of beautiful Amsterdam on generative AI and/or uncertainty quantification 🤗
🎇https://t.co/LCZh4XP6QA
I will be in Montréal until December for my internship with ServiceNow. I will be working on causal discovery from time-series. Get in touch if you are around and want to chat about some **apprentissage profond**
📢📢📢 PhD vacancy alert 📢📢📢
We open several PhD positions supervised by myself and Georgios on #Robot Learning and #Dynamics!
If you have strong #ML and/or #Robotics experience and want to dive into the next big thing in #AI, apply!
Please share!
https://t.co/BOEAOH9F1z
First time organising a Tutorial with an amazing team and am very excited 🎉! The topic is learning from videos, which I think will be the new 'Big' paradigm for new vision foundation models. Come to learn, chat and discuss @eccvconf!
Happy to present this paper accepted @eccvconf: upgrades VideoMAE to use Sinkhorn-Knopp on patch-level embeddings. This moves reconstruction one level up towards more semantic features. Training is simple & stable.