Bethge Lab

Example Talk

Sat, 01 Jun 2030 13:00:00 +0000

Click on the Slides button above to view the built-in slides feature.

Slides can be added in a few ways:

Create slides using Wowchemy’s Slides feature and link using slides parameter in the front matter of the talk file
Upload an existing slide deck to static/ and link using url_slides parameter in the front matter of the talk file
Embed your slides (e.g. Google Slides) or presentation video on this page using shortcodes.

Further event details, including page elements such as image galleries, can be added to the body of this page.

Modeling Saliency Dataset Bias

Mon, 06 Oct 2025 00:00:00 +0000

LLMs on the Line: Data Determines Loss-to-Loss Scaling Laws

Mon, 17 Feb 2025 00:00:00 +0000

Pretraining Frequency Predicts Compositional Generalization of CLIP on Real-World Tasks

Mon, 17 Feb 2025 00:00:00 +0000

CiteME: Can Language Models Accurately Cite Scientific Claims?

Tue, 10 Dec 2024 00:00:00 +0000

How to Merge Your Multimodal Models Over Time?

Mon, 09 Dec 2024 00:00:00 +0000

ONEBench to Test Them All: Sample-Level Benchmarking Over Open-Ended Capabilities

Mon, 09 Dec 2024 00:00:00 +0000

Object segmentation from common fate: Motion energy processing enables human-like zero-shot generalization to random dot stimuli

Sun, 01 Dec 2024 00:00:00 +0000

Reflecting on the State of Rehearsal-free Continual Learning with Pretrained Models

Tue, 22 Oct 2024 00:00:00 +0000

Infinite dSprites for Disentangled Continual Learning: Separating Memory Edits from Generalization

Sun, 20 Oct 2024 00:00:00 +0000

Adaptation Odyssey in LLMs: Why Does Additional Pretraining Sometimes Fail to Improve?

Sat, 12 Oct 2024 09:58:51 +0200

Identifying latent state transition in non-linear dynamical systems

Sat, 12 Oct 2024 09:58:51 +0200

Investigating Continual Pretraining in Large Language Models: Insights and Implications

Sat, 12 Oct 2024 09:58:51 +0200

Modulated Neural ODEs

Sat, 12 Oct 2024 09:58:51 +0200

In Search of Forgotten Domain Generalization

Thu, 10 Oct 2024 00:00:00 +0000

A Practitioner's Guide to Continual Multimodal Pretraining

Mon, 07 Oct 2024 00:00:00 +0000

Efficient Lifelong Model Evaluation in an Era of Rapid Progress

Mon, 07 Oct 2024 00:00:00 +0000

No Zero-Shot Without Exponential Data---Pretraining Concept Frequency Determines Multimodal Model Performance

Mon, 07 Oct 2024 00:00:00 +0000

SuS-X---Training-Free Name-Only Transfer of Vision-Language Models

Mon, 07 Oct 2024 00:00:00 +0000

Visual Data-Type Understanding does not emerge from Scaling Vision-Language Models

Mon, 07 Oct 2024 00:00:00 +0000

A chromatic feature detector in the retina signals visual context changes

Fri, 04 Oct 2024 09:58:51 +0200

Fri, 17 May 2024 09:58:51 +0200

The Entropy Enigma: Success and Failure of Entropy Minimization

Sun, 12 May 2024 00:00:00 +0000

Fri, 10 May 2024 09:58:51 +0200

Deep feature matching vs spatio-temporal energy filtering for robust moving object segmentation

Wed, 01 May 2024 00:00:00 +0000

Scale Learning in Scale-Equivariant Convolutional Networks

Thu, 29 Feb 2024 00:00:00 +0000

Does CLIP's Generalization Performance Mainly Stem from High Train-Test Similarity?

Sat, 14 Oct 2023 00:00:00 +0000

Provable Compositional Generalization for Object-Centric Learning

Mon, 09 Oct 2023 00:00:00 +0000

Compositional Generalization from First Principles

Mon, 10 Jul 2023 00:00:00 +0000

Fri, 07 Jul 2023 09:58:51 +0200

RDumb: A simple approach that questions our progress in continual test-time adaptation

Thu, 08 Jun 2023 00:00:00 +0000

Unsupervised object learning via common fate

Sat, 01 Apr 2023 00:00:00 +0000

Two Day Lab Hackathon 2023

Mon, 13 Feb 2023 00:00:00 +0000

Bethgelab ❤️ ELLIS

Fri, 13 Jan 2023 00:00:00 +0000

AI sciencepreneurship and startups

Sun, 01 Jan 2023 06:00:00 +0000

Machine learning is rapidly expanding the range of skills that can be used for new solutions to relevant problems in the world either by being more scalable or more precise then human labor. We seek to develop a better understanding of how we can develop economically feasible solutions that best address long-term human needs. We spin off and collaborate with startups such as Maddox AI or Black Forest Labs.

Attention in Humans and Machines

Sun, 01 Jan 2023 05:00:00 +0000

Kümmerer, Matthias, and Matthias Bethge. “Predicting visual fixations.” Annual Review of Vision Science 9.1 (2023): 269-291.
Kümmerer, Matthias, et al. “Saliency Beyond Datasets: Overcoming Dataset Biases.” (Coming Soon)
Schwetlick, Lisa, et al. “Potsdam data set of eye movement on natural scenes (DAEMONS).” Frontiers in Psychology 15 (2024): 1389609.
Kümmerer, M., and M. Bethge. “State-of-the-art in human scanpath prediction. ArXiv.” arXiv preprint arXiv:2102.12239(2021).

Modeling brain representations & mechanistic interpretability

Sun, 01 Jan 2023 04:00:00 +0000

Höfling, Larissa, et al. “A chromatic feature detector in the retina signals visual context changes.” eLife 13 (2024): e86860.
Cadena, Santiago A., et al. “Diverse task-driven modeling of macaque V4 reveals functional specialization towards semantic tasks.” PLOS Computational Biology 20.5 (2024): e1012056.
Burg, Max F., et al. “Most discriminative stimuli for functional cell type clustering.” ArXiv (2024).
Sandbrink, Kai J., et al. “Contrasting action and posture coding with hierarchical deep neural network models of proprioception.” Elife 12 (2023): e81499.
Mathis, Alexander, et al. “DeepLabCut: markerless pose estimation of user-defined body parts with deep learning.” Nature neuroscience 21.9 (2018): 1281-1289.
Baden, Tom, et al. “The functional diversity of retinal ganglion cells in the mouse.” Nature 529.7586 (2016): 345-350.

Lifelong compositional, scalable and object-centric learning

Sun, 01 Jan 2023 03:00:00 +0000

Udandarao, Vishaal, et al. “A Practitioner’s Guide to Continual Multimodal Pretraining.” NeurIPS (2024).
Yıldız, Çağatay, et al. “Investigating Continual Pretraining in Large Language Models: Insights and Implications.” arXiv preprint arXiv:2402.17400 (2024).
Wiedemer, Thaddäus, et al. “Compositional generalization from first principles.” NeurIPS (2023).
Dziadzio, Sebastian, et al. “Disentangled Continual Learning: Separating Memory Edits from Model Updates.” CoLLAs (2024).
Wiedemer, Thaddäus, et al. “Provable Compositional Generalization for Object-Centric Learning.” arXiv preprint arXiv:2310.05327 (2023).
Tangemann, Matthias, et al. “Object segmentation from common fate: Motion energy processing enables human-like zero-shot generalization to random dot stimuli.” NeurIPS (2024)

Language Model Agents

Sun, 01 Jan 2023 02:00:00 +0000

Press, Ori, et al. “CiteME: Can Language Models Accurately Cite Scientific Claims?.” NeurIPS 2024.
Sinha, Shiven et al. “Wu’s Method can Boost Symbolic AI to Rival Silver Medalists and AlphaGeometry to Outperform Gold Medalists at IMO Geometry”, NeurIPS Math-in-AI Workshop (2024)
Akata, Elif, et al. “Playing repeated games with large language models.” arXiv preprint arXiv:2305.16867 (2023).

Open-ended model evaluation & benchmarking

Sun, 01 Jan 2023 01:00:00 +0000

Prabhu, Ameya, et al. “Efficient Lifelong Model Evaluation in an Era of Rapid Progress.” NeurIPS 2024.
Ghosh, Adhiraj, et al. “Democratizing Evaluation with Infinity-Benchmarks: Sample-Level Heterogeneous Testing Over Arbitrary Capabilities”, Arxiv
Kümmerer, Matthias, et al. “Saliency Beyond Datasets: Overcoming Dataset Biases.” (Coming Soon)
Udandarao, Vishaal, et al. “No ‘zero-shot’ without exponential data: Pretraining concept frequency determines multimodal model performance.” NeurIPS 2024.
Berens, Philipp, et al. “Community-based benchmarking improves spike rate inference from two-photon calcium imaging data.” PLoS computational biology 14.5 (2018): e1006157.

DeepGaze vs SceneWalk: what can DNNs and biological scan path models teach each other?

Mon, 05 Dec 2022 00:00:00 +0000

The bittersweet lesson: data-rich models narrow the behavioural gap to human vision

Mon, 05 Dec 2022 00:00:00 +0000

Using the DeepGaze III model to decompose spatial and dynamic contributions to fixation placement over time

Mon, 05 Dec 2022 00:00:00 +0000

Disentanglement and generalization under correlation shifts

Mon, 28 Nov 2022 00:00:00 +0000

Common fate based object learning in machines and humans

Sun, 01 May 2022 00:00:00 +0000

Deepgaze iii: Modeling free-viewing human scanpaths with deep learning

Wed, 06 Apr 2022 00:00:00 +0000

Digital twin reveals combinatorial code of non-linear computations in the mouse primary visual cortex

Thu, 10 Feb 2022 00:00:00 +0000

Robust deep learning object recognition models rely on low frequency information in natural images

Wed, 02 Feb 2022 00:00:00 +0000

Efficient coding of natural scenes improves neural system identification

Tue, 11 Jan 2022 00:00:00 +0000

Diverse task-driven modeling of macaque V4 reveals functional specialization towards semantic tasks

Sat, 01 Jan 2022 00:00:00 +0000

ImageNet-D: A new challenging robustness dataset inspired by domain adaptation

Sat, 01 Jan 2022 00:00:00 +0000

Semantic object-scene inconsisten

Sat, 01 Jan 2022 00:00:00 +0000

How Well do Feature Visualizations Support Causal Understanding of CNN Activations?

Mon, 06 Dec 2021 00:00:00 +0000

Partial success in closing the gap between human and machine vision

Mon, 06 Dec 2021 00:00:00 +0000

Contextualised meaning maps do not predict how semantic object-context inconsistencies change human gaze behaviour

Wed, 01 Dec 2021 00:00:00 +0000

2.1 Decision-Based Adversarial Attacks

Fri, 05 Nov 2021 00:00:00 +0000

If your data distribution shifts, use self-learning

Wed, 29 Sep 2021 00:00:00 +0000

New enhancements to the DeepGaze models for a better understanding of human scanpaths

Mon, 27 Sep 2021 00:00:00 +0000

Object-context inconsistencies affect gaze behavior differently than predicted by contextualized meaning maps

Mon, 27 Sep 2021 00:00:00 +0000

There is no evidence that meaning maps capture semantic information relevant to gaze guidance: Reply to Henderson, Hayes, Peacock, and Rehrig (2021)

Wed, 01 Sep 2021 00:00:00 +0000

Visual representation learning does not generalize strongly within the same domain

Sat, 17 Jul 2021 00:00:00 +0000

Contrastive learning inverts the data generating process

Thu, 01 Jul 2021 00:00:00 +0000

Learning divisive normalization in primary visual cortex

Mon, 07 Jun 2021 00:00:00 +0000

Semantic object-scene inconsistencies affect eye movements, but not in the way predicted by contextualized meaning maps

Tue, 04 May 2021 00:00:00 +0000

Adapting imagenet-scale models to complex distribution shifts with self-learning

Tue, 27 Apr 2021 00:00:00 +0000

Selected Inductive Biases in Neural Networks To Generalize Beyond the Training Domain

Mon, 12 Apr 2021 00:00:00 +0000

Five points to check when comparing visual perception in humans and machines

Mon, 01 Mar 2021 00:00:00 +0000

State-of-the-art in human scanpath prediction

Wed, 24 Feb 2021 00:00:00 +0000

A diverse task-driven characterization of early and mid-level representations of the primate ventral stream