Matthias Kümmerer

521 posts

Matthias Kümmerer

Matthias Kümmerer

@matthiaskue

Neuroscience, Machine Learning and Classical Music. Postdoc @bethgelab. Founding Member of @grammofy.

Katılım Ekim 2013
330 Takip Edilen472 Takipçiler
Matthias Kümmerer
Matthias Kümmerer@matthiaskue·
I'm very excited to present our paper together with @fededagos here at #NeurIPS2025 today at 4:30pm, poster no 2103. Come by if you want to see how DNNs can help build better mechanistic models of human behaviour!
Federico D'Agostino@fededagos

🚨 New paper at #NeurIPS2025! A systematic fixation-level comparison of a performance-optimized DNN scanpath model and a mechanistic cognitive model reveals behaviourally relevant mechanisms that can be added to the mechanistic model to substantially improve performance. 🧵👇

English
0
1
3
360
Matthias Kümmerer retweetledi
Federico D'Agostino
Federico D'Agostino@fededagos·
🚨 New paper at #NeurIPS2025! A systematic fixation-level comparison of a performance-optimized DNN scanpath model and a mechanistic cognitive model reveals behaviourally relevant mechanisms that can be added to the mechanistic model to substantially improve performance. 🧵👇
Federico D'Agostino tweet media
English
1
4
8
1.9K
Matthias Kümmerer retweetledi
Vishaal Udandarao✈️ICLR’26
Vishaal Udandarao✈️ICLR’26@vishaal_urao·
🚀New Paper arxiv.org/abs/2511.20643 Most data-curation for contrastive VLMs is offline and concept-agnostic! We introduce online, concept-aware batch sampling (CABS), enabling flexible and targeted curation, yielding ⬆️boosts over IID sampling (+7-9% on ImageNet + ret)! 🧵👇
Vishaal Udandarao✈️ICLR’26 tweet media
English
1
4
22
976
Matthias Kümmerer
Matthias Kümmerer@matthiaskue·
Our model sets a new SOTA on multiple benchmark datasets for generalization, adapation and full training.
English
0
0
0
22
Matthias Kümmerer
Matthias Kümmerer@matthiaskue·
We found that less than 20 interpretable parameters are responsible for most of the performance drop on unseen datasets and that they can be adapted in a very data efficient way.
Matthias Kümmerer tweet media
English
1
0
0
27
Matthias Kümmerer
Matthias Kümmerer@matthiaskue·
I'm happy to be here at @ICCVConference and looking forward to meeting many interesting people! I will present our new paper on Modeling Saliency Dataset Bias for better predictions on new datasets: youtu.be/gOkVY4cNVLo
YouTube video
YouTube
English
1
0
1
63
Matthias Kümmerer
Matthias Kümmerer@matthiaskue·
@brialong Sounds exciting! Do you happen to have a timecode? The video is quite long and I couldn't find a table of content.
English
0
0
1
58
Matthias Kümmerer retweetledi
Vishaal Udandarao✈️ICLR’26
Vishaal Udandarao✈️ICLR’26@vishaal_urao·
🚀New Preprint Alert! 📊Exploring the notion of "Zero-Shot" Generalization in Foundation Models. Is it all just a myth? Our latest preprint dives deep. Check it out!🔍 arxiv.org/abs/2404.04125
Vishaal Udandarao✈️ICLR’26 tweet media
English
8
31
158
32.6K
Matthias Kümmerer retweetledi
Ori Press
Ori Press@ori_press·
Interested in deploying computer vision models in the real world? You're probably going to want to continuously adapt to your data. We show that a simple baseline approach beats all current continuously adapting methods. ⬇️🧵 1/6
English
2
9
37
6.2K
Matthias Kümmerer
Matthias Kümmerer@matthiaskue·
I'm looking forward to giving my talk "Modeling Internal State Changes in Free-Viewing and Visual Search Scanpaths With Gain Control in DeepGaze III" at 12:00pm @VSSMtg . Come by if you're interested!
Matthias Kümmerer tweet mediaMatthias Kümmerer tweet media
English
0
2
24
2K
Gavin Buckingham
Gavin Buckingham@DrGBuckingham·
I don't edit papers my direct trainees send me with tracked changes - I edit live in front of them in a meeting. Why? 1. forces you to find time to do it 2. makes you accountable for the quality of your edits 3. makes the edits a discussion 4. articulate why you make an edit
English
32
31
554
0
Matthias Kümmerer
Matthias Kümmerer@matthiaskue·
Come see Lisa's talk about our joint project on bringing mechanistic scanpath models and deep learning scanpath models together! #VSS2022
English
0
0
9
0

Keşfet

@fededagos @ICCVConference @brialong @TrackingActions @stes_io @ELLISforEurope @bethgelab @EDNE_EPFL