Matthias De Lange

111 posts

Matthias De Lange

Matthias De Lange

@MatthiasDeLange

Senior AI researcher @Techwolf | PhD adaptive machine learning | ex-@MetaAI | Chair CLVision @CVPR

Leuven Katılım Nisan 2017
284 Takip Edilen243 Takipçiler
Sabitlenmiş Tweet
Matthias De Lange
Matthias De Lange@MatthiasDeLange·
Looking forward to talk about our new ICLR spotlight paper 🔥 Introducing continual evaluation, we consistently find a 'Stability Gap' for replay and regularization methods! 🗨️ Come join the talk March 30th, 📜and check out the paper for a sneak peek: openreview.net/forum?id=Zy350…
ContinualAI@ContinualAI

🔥Ready for the next ContinualAI Seminar!?🔥 Join us next Thursday (March 30th at 9am UTC) for "Bridging the Gap: Evaluating and Applying Continual Learning" with @MatthiasDeLange @camgbus @soroorh1 Receive your invitation (w/ calendar link) by visiting: forms.gle/QKkbzHaEwWY44B…

English
2
3
17
1.4K
Matthias De Lange retweetledi
Gido van de Ven
Gido van de Ven@GMvandeVen·
An exciting and fully-funded PhD position is available to work on continual learning with Tinne Tuytelaars and me. The focus is on conceptual and theoretical understanding, not on chasing state-of-the-art. If this matches your interest, please consider applying!
Tuytelaars Lab@tuytelaarslab

A PhD position on “continual learning with deep neural networks” is available in our lab! Apply before 17 January 2024. kuleuven.be/personeel/jobs…

English
0
8
29
2.2K
Matthias De Lange
Matthias De Lange@MatthiasDeLange·
In my latest blogpost with #Radix, I'll cover: 🔥 What is Adaptive AI and how to leverage it for your business? 🚀 How to create a personalized Large Language Model (LLM), tailored to your company's jargon. 🎯 The choice between Continual Learning or Quick Adaptation.
English
1
0
0
83
Matthias De Lange
Matthias De Lange@MatthiasDeLange·
While AI is not designed to change, ironically, change is the one constant in our world🤔 Stanford's study shows that even GPT-4's performance decreased over time, stressing that outdated AI also affects the industry🚨 Is Adaptive AI the solution? matthias-de-lange.medium.com/adaptive-ai-cr…
English
1
0
6
295
Matthias De Lange
Matthias De Lange@MatthiasDeLange·
Our Multiscale Video Pretraining (MVP) is a new self-supervised approach to predict long-term activity forecasting! We attain SOTA due to predicting contextualized representations of future video clips over multiple timescales 🔥 Checkout the paper ⬇️ #MetaAI #MVP
AK@_akhaliq

Multiscale Video Pretraining for Long-Term Activity Forecasting paper page: huggingface.co/papers/2307.12… Long-term activity forecasting is an especially challenging research problem because it requires understanding the temporal relationships between observed actions, as well as the variability and complexity of human activities. Despite relying on strong supervision via expensive human annotations, state-of-the-art forecasting approaches often generalize poorly to unseen data. To alleviate this issue, we propose Multiscale Video Pretraining (MVP), a novel self-supervised pretraining approach that learns robust representations for forecasting by learning to predict contextualized representations of future video clips over multiple timescales. MVP is based on our observation that actions in videos have a multiscale nature, where atomic actions typically occur at a short timescale and more complex actions may span longer timescales. We compare MVP to state-of-the-art self-supervised video learning approaches on downstream long-term forecasting tasks including long-term action anticipation and video summary prediction. Our comprehensive experiments across the Ego4D and Epic-Kitchens-55/100 datasets demonstrate that MVP out-performs state-of-the-art methods by significant margins. Notably, MVP obtains a relative performance gain of over 20% accuracy in video summary forecasting over existing methods.

English
0
0
2
289
Matthias De Lange retweetledi
Vincenzo Lomonaco
Vincenzo Lomonaco@v_lomonaco·
Hi everyone, I'm looking for prospective PhDs interested in @ContinualAI and related topics, please send me an email if you want to learn more!
English
1
12
43
9K
Matthias De Lange
Matthias De Lange@MatthiasDeLange·
The video naturally contains long-tailed action distributions (with both sudden and gradual shifts) and large-scale classification (of over 2.7k actions), providing a perfect testbed for real-world online continual learning and personalization.
English
1
0
1
86
Matthias De Lange
Matthias De Lange@MatthiasDeLange·
I’m super excited to finally share the research I did at #MetaAI! Working on #ContinualLearning during my Ph.D. and Meta’s focus on head-mounted devices made a great synergy.🔥 Why is that? Let's find out below ⬇️ (or read the paper📃: lnkd.in/erAuadtY )
English
1
8
29
3.2K
Matthias De Lange retweetledi
Gido van de Ven
Gido van de Ven@GMvandeVen·
It’s time for the CLVision workshop at #CVPR 2023! Starting soon in room East 2, or join virtually on Zoom (needs CVPR registration). Looking forward to great speakers, >30 contributed papers, a panel discussion and reports from an exciting challenge! sites.google.com/view/clvision2…
English
0
8
26
5.4K
Matthias De Lange retweetledi
ContinualAI
ContinualAI@ContinualAI·
Missed our March seminar/meet-up? No problem! 🚀 Catch up now on YouTube: youtu.be/T2IYBSyug6w
YouTube video
YouTube
English
0
4
9
970
Matthias De Lange retweetledi
Ali Diba
Ali Diba@AliDiba67·
I'm hiring for a few computer vision engineer and internship positions (Not Remote) in Belgium. If you have hands-on experience in CV/ML, especially computer vision on-edge, please send me your CV. #computervision #hiring #edgeai
English
1
4
11
1.2K
Matthias De Lange
Matthias De Lange@MatthiasDeLange·
Looking forward to talk about our new ICLR spotlight paper 🔥 Introducing continual evaluation, we consistently find a 'Stability Gap' for replay and regularization methods! 🗨️ Come join the talk March 30th, 📜and check out the paper for a sneak peek: openreview.net/forum?id=Zy350…
ContinualAI@ContinualAI

🔥Ready for the next ContinualAI Seminar!?🔥 Join us next Thursday (March 30th at 9am UTC) for "Bridging the Gap: Evaluating and Applying Continual Learning" with @MatthiasDeLange @camgbus @soroorh1 Receive your invitation (w/ calendar link) by visiting: forms.gle/QKkbzHaEwWY44B…

English
2
3
17
1.4K
Matthias De Lange retweetledi
Gido van de Ven
Gido van de Ven@GMvandeVen·
The paper submission deadline for the CLVision workshop has been extended to March 13th. There is still time to submit your (archival or non-archival) contribution! #CVPR #continuallearning
Gido van de Ven@GMvandeVen

Details for the Continual Learning #CLVision workshop at #CVPR2023 are announced! 🗨️ exciting line-up of speakers 📜 call for papers (deadline 7 March), with archival and non-archival track 🔥 a challenge (coming soon) 📅 date of workshop: 18 June sites.google.com/view/clvision2…

English
0
5
9
1.4K
Matthias De Lange
Matthias De Lange@MatthiasDeLange·
You want to be part of the exciting Continual Learning #CLVISION workshop at @CVPR? 📜 Submit your work on continual/few-shot/reinforcement learning. Also non-vision papers! (7 March) 🔥 participate in the challenge for amazing prizes from our sponsors: sites.google.com/view/clvision2…
Gido van de Ven@GMvandeVen

Details for the Continual Learning #CLVision workshop at #CVPR2023 are announced! 🗨️ exciting line-up of speakers 📜 call for papers (deadline 7 March), with archival and non-archival track 🔥 a challenge (coming soon) 📅 date of workshop: 18 June sites.google.com/view/clvision2…

English
0
5
20
7K
Matthias De Lange retweetledi
Gido van de Ven
Gido van de Ven@GMvandeVen·
Details for the Continual Learning #CLVision workshop at #CVPR2023 are announced! 🗨️ exciting line-up of speakers 📜 call for papers (deadline 7 March), with archival and non-archival track 🔥 a challenge (coming soon) 📅 date of workshop: 18 June sites.google.com/view/clvision2…
English
1
16
45
21.6K