Wieland Brendel

Who Guards the Guardians? The Challenges of Evaluating Identifiability of Learned Representations UAI 2026 Cross-Entropy Is All You Need To Invert the Data Generating Process ICLR 2025 Identifiable Exchangeable Mechanisms for Causal Structure and Representation Learning ICLR 2025 InfoNCE: Identifying the Gap Between Theory and Practice AISTATS 2025 In Search of Forgotten Domain Generalization ICLR 2025 Interaction Asymmetry: A General Principle for Learning Composable Abstractions ICLR 2025 LAION-C: An Out-of-Distribution Benchmark for Web-Scale Vision Models ICML 2025 LLMs on the Line: Data Determines Loss-to-Loss Scaling Laws ICML 2025 Position: An Empirically Grounded Identifiability Theory Will Accelerate Self Supervised Learning Research ICML 2025 VGGSounder: Audio-Visual Evaluations for Foundation Models ICCV 2025 An Interventional Perspective on Identifiability in Gaussian LTI Systems with Independent Component Analysis CLeaR 2024 Does CLIP’s generalization performance mainly stem from high train-test similarity? ICLR 2024 Don’t trust your eyes: on the (un)reliability of feature visualizations ICML 2024 Effective pruning of web-scale datasets based on complexity of concept clusters ICLR 2024 Measuring Per-Unit Interpretability at Scale Without Humans NeurIPS 2024 Position: Understanding LLMs Requires More Than Statistical Generalization ICML 2024 Provable Compositional Generalization for Object-Centric Learning ICLR 2024 Rule Extrapolation in Language Modeling: A Study of Compositional Generalization on OOD Prompts NeurIPS 2024 Compositional Generalization from First Principles NeurIPS 2023 Provably Learning Object-Centric Representations ICML 2023 Scale Alone Does not Improve Mechanistic Interpretability in Vision Models NeurIPS 2023 Embrace the Gap: VAEs Perform Independent Mechanism Analysis NeurIPS 2022 Increasing Confidence in Adversarial Robustness Evaluations NeurIPS 2022 Visual Representation Learning Does Not Generalize Strongly Within the Same Domain ICLR 2022 Benchmarking Unsupervised Object Representations for Video Sequences JMLR 2021 Contrastive Learning Inverts the Data Generating Process ICML 2021 Exemplary Natural Images Explain CNN Activations Better than State-of-the-Art Feature Visualization ICLR 2021 Fast Minimum-norm Adversarial Attacks through Adaptive Norm Constraints NeurIPS 2021 How Well do Feature Visualizations Support Causal Understanding of CNN Activations? NeurIPS 2021 Partial success in closing the gap between human and machine vision NeurIPS 2021 Self-Supervised Learning with Data Augmentations Provably Isolates Content from Style NeurIPS 2021 Towards Nonlinear Disentanglement in Natural Data with Temporal Sparse Coding ICLR 2021 A Simple Way to Make Neural Networks Robust Against Diverse Image Corruptions ECCV 2020 Improving robustness against common corruptions by covariate shift adaptation NeurIPS 2020 On Adaptive Attacks to Adversarial Example Defenses NeurIPS 2020 Accurate, reliable and fast robustness evaluation NeurIPS 2019 Approximating CNNs with Bag-of-local-Features models works surprisingly well on ImageNet ICLR 2019 ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness ICLR 2019 Learning from brains how to regularize machines NeurIPS 2019 Towards the first adversarially robust neural network model on MNIST ICLR 2019 Unsupervised learning of an efficient short-term memory network NeurIPS 2014 Demixed Principal Component Analysis NeurIPS 2011