arXiv subjects

Markus Kollmann

Publications and source records attributed to Markus Kollmann.

11 recordsLinked to original sources

arXiv2026

Diffusion Models Memorize in Training -- and Generalize in Inference

Diffusion models generalize well in practice. However, an optimal diffusion model fully memorizes the training data and therefore fails to generalize, raising the question of what induces generalization in a real diffusion model. We show that, despite generalizing at the sample level, diffusion models progressively overfit the denoising training objective and thereby create a generalization gap between the performance on validation and training samples. This gap is most pronounced at intermediate noise levels. Using a fully analytic error-prone toy model, we trace the factors affecting the generalization gap. We find that the optimal denoising flow field localizes sharply around training points, but the model error suppresses the exact recall of training points, yielding a smooth, generalizing flow field. Finally, we find that the generalization gap observed in training does not translate to inference, which would result in a strong similarity between generated samples and training samples. This is because the intermediate states of sampling trajectories are sufficiently far from the distribution of noisy training samples the model is trained on. Together, these findings reveal a novel picture of how diffusion models generalize: the flow field generalizes through model error, which moves sampling trajectories outside the domain of noisy training samples and thereby naturally prevents overfitting.

cs.LG↗

arXiv2025

ClusterMine: Robust Label-Free Visual Out-Of-Distribution Detection via Concept Mining from Text Corpora

Large-scale visual out-of-distribution (OOD) detection has witnessed remarkable progress by leveraging vision-language models such as CLIP. However, a significant limitation of current methods is their reliance on a pre-defined set of in-distribution (ID) ground-truth label names (positives). These fixed label names can be unavailable, unreliable at scale, or become less relevant due to in-distribution shifts after deployment. Towards truly unsupervised OOD detection, we utilize widely available text corpora for positive label mining, bypassing the need for positives. In this paper, we utilize widely available text corpora for positive label mining under a general concept mining paradigm. Within this framework, we propose ClusterMine, a novel positive label mining method. ClusterMine is the first method to achieve state-of-the-art OOD detection performance without access to positive labels. It extracts positive concepts from a large text corpus by combining visual-only sample consistency (via clustering) and zero-shot image-text consistency. Our experimental study reveals that ClusterMine is scalable across a plethora of CLIP models and achieves state-of-the-art robustness to covariate in-distribution shifts. The code is available at https://github.com/HHU-MMBS/clustermine_wacv_official.

cs.CV↗

arXiv2024

Guiding a diffusion model using sliding windows

Guidance is a widely used technique for diffusion models to enhance sample quality. Technically, guidance is realised by using an auxiliary model that generalises more broadly than the primary model. Using a 2D toy example, we first show that it is highly beneficial when the auxiliary model exhibits similar but stronger generalisation errors than the primary model. Based on this insight, we introduce \emph{masked sliding window guidance (M-SWG)}, a novel, training-free method. M-SWG upweights long-range spatial dependencies by guiding the primary model with itself by selectively restricting its receptive field. M-SWG requires neither access to model weights from previous iterations, additional training, nor class conditioning. M-SWG achieves a superior Inception score (IS) compared to previous state-of-the-art training-free approaches, without introducing sample oversaturation. In conjunction with existing guidance methods, M-SWG reaches state-of-the-art Frechet DINOv2 distance on ImageNet using EDM2-XXL and DiT-XL. The code is available at https://github.com/HHU-MMBS/swg_bmvc2025_official.

cs.CV↗

arXiv2024

Scaling Up Deep Clustering Methods Beyond ImageNet-1K

Deep image clustering methods are typically evaluated on small-scale balanced classification datasets while feature-based $k$-means has been applied on proprietary billion-scale datasets. In this work, we explore the performance of feature-based deep clustering approaches on large-scale benchmarks whilst disentangling the impact of the following data-related factors: i) class imbalance, ii) class granularity, iii) easy-to-recognize classes, and iv) the ability to capture multiple classes. Consequently, we develop multiple new benchmarks based on ImageNet21K. Our experimental analysis reveals that feature-based $k$-means is often unfairly evaluated on balanced datasets. However, deep clustering methods outperform $k$-means across most large-scale benchmarks. Interestingly, $k$-means underperforms on easy-to-classify benchmarks by large margins. The performance gap, however, diminishes on the highest data regimes such as ImageNet21K. Finally, we find that non-primary cluster predictions capture meaningful classes (i.e. coarser classes).

cs.CV↗

arXiv2024

Rethinking cluster-conditioned diffusion models for label-free image synthesis

Diffusion-based image generation models can enhance image quality when conditioned on ground truth labels. Here, we conduct a comprehensive experimental study on image-level conditioning for diffusion models using cluster assignments. We investigate how individual clustering determinants, such as the number of clusters and the clustering method, impact image synthesis across three different datasets. Given the optimal number of clusters with respect to image synthesis, we show that cluster-conditioning can achieve state-of-the-art performance, with an FID of 1.67 for CIFAR10 and 2.17 for CIFAR100, along with a strong increase in training sample efficiency. We further propose a novel empirical method to estimate an upper bound for the optimal number of clusters. Unlike existing approaches, we find no significant association between clustering performance and the corresponding cluster-conditional FID scores. The code is available at https://github.com/HHU-MMBS/cedm-official-wavc2025.

cs.CV↗

arXiv2023

Exploring the Limits of Deep Image Clustering using Pretrained Models

We present a general methodology that learns to classify images without labels by leveraging pretrained feature extractors. Our approach involves self-distillation training of clustering heads based on the fact that nearest neighbours in the pretrained feature space are likely to share the same label. We propose a novel objective that learns associations between image features by introducing a variant of pointwise mutual information together with instance weighting. We demonstrate that the proposed objective is able to attenuate the effect of false positive pairs while efficiently exploiting the structure in the pretrained feature space. As a result, we improve the clustering accuracy over $k$-means on $17$ different pretrained models by $6.1$\% and $12.2$\% on ImageNet and CIFAR100, respectively. Finally, using self-supervised vision transformers, we achieve a clustering accuracy of $61.6$\% on ImageNet. The code is available at https://github.com/HHU-MMBS/TEMI-official-BMVC2023.

cs.CV↗

arXiv2023

Adapting Contrastive Language-Image Pretrained (CLIP) Models for Out-of-Distribution Detection

We present a comprehensive experimental study on pretrained feature extractors for visual out-of-distribution (OOD) detection, focusing on adapting contrastive language-image pretrained (CLIP) models. Without fine-tuning on the training data, we are able to establish a positive correlation ($R^2\geq0.92$) between in-distribution classification and unsupervised OOD detection for CLIP models in $4$ benchmarks. We further propose a new simple and scalable method called \textit{pseudo-label probing} (PLP) that adapts vision-language models for OOD detection. Given a set of label names of the training set, PLP trains a linear layer using the pseudo-labels derived from the text encoder of CLIP. To test the OOD detection robustness of pretrained models, we develop a novel feature-based adversarial OOD data manipulation approach to create adversarial samples. Intriguingly, we show that (i) PLP outperforms the previous state-of-the-art \citep{ming2022mcm} on all $5$ large-scale benchmarks based on ImageNet, specifically by an average AUROC gain of 3.4\% using the largest CLIP model (ViT-G), (ii) we show that linear probing outperforms fine-tuning by large margins for CLIP architectures (i.e. CLIP ViT-H achieves a mean gain of 7.3\% AUROC on average on all ImageNet-based benchmarks), and (iii) billion-parameter CLIP models still fail at detecting adversarially manipulated OOD images. The code and adversarially created datasets will be made publicly available.

cs.CV↗

arXiv2022

Self-Supervised Anomaly Detection by Self-Distillation and Negative Sampling

Detecting whether examples belong to a given in-distribution or are Out-Of-Distribution (OOD) requires identifying features specific to the in-distribution. In the absence of labels, these features can be learned by self-supervised techniques under the generic assumption that the most abstract features are those which are statistically most over-represented in comparison to other distributions from the same domain. In this work, we show that self-distillation of the in-distribution training set together with contrasting against negative examples derived from shifting transformation of auxiliary data strongly improves OOD detection. We find that this improvement depends on how the negative samples are generated. In particular, we observe that by leveraging negative samples, which keep the statistics of low-level features while changing the high-level semantics, higher average detection performance is obtained. Furthermore, good negative sampling strategies can be identified from the sensitivity of the OOD detection score. The efficiency of our approach is demonstrated across a diverse range of OOD detection problems, setting new benchmarks for unsupervised OOD detection in the visual domain.

cs.AI↗

arXiv2021

Unsupervised Anomaly Detection From Semantic Similarity Scores

Classifying samples as in-distribution or out-of-distribution (OOD) is a challenging problem of anomaly detection and a strong test of the generalisation power for models of the in-distribution. In this paper, we present a simple and generic framework, {\it SemSAD}, that makes use of a semantic similarity score to carry out anomaly detection. The idea is to first find for any test example the semantically closest examples in the training set, where the semantic relation between examples is quantified by the cosine similarity between feature vectors that leave semantics unchanged under transformations, such as geometric transformations (images), time shifts (audio signals), and synonymous word substitutions (text). A trained discriminator is then used to classify a test example as OOD if the semantic similarity to its nearest neighbours is significantly lower than the corresponding similarity for test examples from the in-distribution. We are able to outperform previous approaches for anomaly, novelty, or out-of-distribution detection in the visual domain by a large margin. In particular, we obtain AUROC values close to one for the challenging task of detecting examples from CIFAR-10 as out-of-distribution given CIFAR-100 as in-distribution, without making use of label information.

cs.LG↗

arXiv2007

Induction level determines signature of gene expression noise in cellular systems

Noise in gene expression, either due to inherent stochasticity or to varying inter- and intracellular environment, can generate significant cell-to-cell variability of protein levels in clonal populations. We present a theoretical framework, based on stochastic processes, to quantify the different sources of gene expression noise taking cell division explicitly into account. Analytical, time-dependent solutions for the noise contributions arising from the major steps involved in protein synthesis are derived. The analysis shows that the induction level of the activator or transcription factor is crucial for the characteristic signature of the dominant source of gene expression noise and thus bridges the gap between seemingly contradictory experimental results. Furthermore, on the basis of experimentally measured cell distributions, our simulations suggest that transcription factor binding and promoter activation can be modelled independently of each other with sufficient accuracy.

q-bio.OT↗

arXiv2002

Single-File Diffusion of Atomic and Colloidal Systems: Asymptotic Laws

In this work we present a general derivation of the non-Fickian behavior for the self-diffusion of identically interacting particle systems with excluded mutual passage. We show that the conditional probability distribution of finding a particle at position $x_{t}$ after time $t$, when the particle was located at $x_{0}$ at $t=0$, follows a Gaussian distribution in the long-time limit, with variance $2W(t)\sim t^{1/2}$ for overdamped systems and with variance $2W(t)\sim t$ for classical systems. The asymptotic behavior of the mean-squared displacement, $W(t)$, is shown to be independent of the nature of interactions for homogeneous systems in the fluid state. Moreover, the long-time behavior of self-diffusion is determined by short-time and large scale collective density fluctuations.

cond-mat.stat-mech↗