arXiv subjects

Marcel Kollovieh

Publications and source records attributed to Marcel Kollovieh.

16 recordsLinked to original sources

arXiv2026

Generative Modeling with Bayesian Sample Inference

We present a novel view of diffusion-like generative modeling from the perspective of iterative Gaussian posterior inference. By treating the generated sample as an unknown variable, we formulate the sampling process in the language of Bayesian probability: at each step, a model predicts the unknown sample from our current belief state and we compute a posterior belief from that prediction. Based on this formulation, we propose the generative model Bayesian Sample Inference (BSI). In addition to a rigorous theoretical analysis, we show that our perspective includes Bayesian Flow Networks (BFNs) and that BSI and BFNs represent the two opposite ends of a range of isotropic hyper-priors over the initial belief state. In our experiments, we demonstrate that BSI improves sample quality over both BFNs and the closely related Variational Diffusion Models on ImageNet32 and ImageNet64, while achieving equivalent log-likelihoods on both datasets.

cs.LG↗

arXiv2026

Byte Pair Encoding for Efficient Time Series Forecasting

Existing time series tokenization methods predominantly encode a constant number of samples into individual tokens. This inflexible approach can generate excessive tokens for even simple patterns like extended constant values, resulting in substantial computational overhead. Inspired by the success of byte pair encoding, we propose the first pattern-centric tokenization scheme for time series analysis. Based on a discrete vocabulary of frequent motifs, our method merges samples with underlying patterns into tokens, compressing time series adaptively. Exploiting our finite set of motifs and the continuous properties of time series, we further introduce conditional decoding as a lightweight yet powerful post-hoc optimization method, which requires no gradient computation and adds no computational overhead. On recent time series foundation models, our motif-based tokenization improves forecasting performance by 40% and boosts efficiency by 2314% on average. Conditional decoding further reduces MSE by up to 48%. In an extensive analysis, we demonstrate the adaptiveness of our tokenization to diverse temporal patterns, its generalization to unseen data, and its meaningful token representations capturing distinct time series properties, including statistical moments and trends.

cs.LG↗

arXiv2026

Energy-Weighted Flow Matching: Unlocking Continuous Normalizing Flows for Efficient and Scalable Boltzmann Sampling

Sampling from unnormalized target distributions, e.g.\ Boltzmann distributions $μ_{\text{target}}(x) \propto \exp(-E(x)/T)$, is fundamental to many scientific applications yet computationally challenging due to complex, high-dimensional energy landscapes. Existing approaches applying modern generative models to Boltzmann distributions either require large datasets of samples drawn from the target distribution or, when using only energy evaluations for training, cannot efficiently leverage the expressivity of advanced architectures like continuous normalizing flows that have shown promise for molecular sampling. To address these shortcomings, we introduce Energy-Weighted Flow Matching (EWFM), a novel training objective enabling continuous normalizing flows to model Boltzmann distributions using only energy function evaluations. Our objective reformulates conditional flow matching via importance sampling, allowing training with samples from arbitrary proposal distributions. Based on this objective, we develop two algorithms: iterative EWFM (iEWFM), which progressively refines proposals through iterative training, and annealed EWFM (aEWFM), which additionally incorporates temperature annealing for challenging energy landscapes. On benchmark systems, including challenging 55-particle Lennard-Jones clusters, our algorithms demonstrate sample quality competitive with established energy-only methods while requiring up to three orders of magnitude fewer energy evaluations.

stat.ML↗

arXiv2026

Interpolating Discrete Diffusion Models with Controllable Resampling

Discrete diffusion models form a powerful class of generative models across diverse domains, including text and graphs. However, existing approaches face fundamental limitations. Masked diffusion models suffer from irreversible errors due to early unmasking, while uniform diffusion models, despite enabling self-correction, often yield low-quality samples due to their strong reliance on intermediate latent states. We introduce IDDM, an Interpolating Discrete Diffusion Model, that improves diffusion by reducing dependence on intermediate latent states. Central to IDDM is a controllable resampling mechanism that partially resets probability mass to the marginal distribution, mitigating error accumulation and enabling more effective token corrections. IDDM specifies a generative process whose transitions interpolate between staying at the current state, resampling from a prior, and flipping toward the target state, while enforcing marginal consistency and fully decoupling training from inference. We benchmark our model against state-of-the-art discrete diffusion models across molecular graph generation as well as text generation tasks, demonstrating competitive performance.

cs.LG↗

arXiv2026

Discrete Bayesian Sample Inference for Graph Generation

Generating graph-structured data is crucial in applications such as molecular generation, knowledge graphs, and network analysis. However, their discrete, unordered nature makes them difficult for traditional generative models, leading to the rise of discrete diffusion and flow matching models. In this work, we introduce GraphBSI, a novel one-shot graph generative model based on Bayesian Sample Inference (BSI). Instead of evolving samples directly, GraphBSI iteratively refines a belief over graphs in the continuous space of distribution parameters, naturally handling discrete structures. Further, we state BSI as a stochastic differential equation (SDE) and derive a noise-controlled family of SDEs that preserves the marginal distributions via an approximation of the score function. Our theoretical analysis further reveals the connection to Bayesian Flow Networks and Diffusion models. Finally, in our empirical evaluation, we demonstrate state-of-the-art performance on molecular and synthetic graph generation, outperforming existing one-shot graph generative models on the standard benchmarks Moses and GuacaMol.

cs.LG↗

arXiv2026

Edit-Based Flow Matching for Temporal Point Processes

Temporal point processes (TPPs) are a fundamental tool for modeling event sequences in continuous time, but most existing approaches rely on autoregressive parameterizations that are limited by their sequential sampling. Recent non-autoregressive, diffusion-style models mitigate these issues by jointly interpolating between noise and data through event insertions and deletions in a discrete Markov chain. In this work, we generalize this perspective and introduce an Edit Flow process for TPPs that transports noise to data via insert, delete, and substitute edit operations. By learning the instantaneous edit rates within a continuous-time Markov chain framework, we attain a flexible and efficient model that effectively reduces the total number of necessary edit operations during generation. Empirical results demonstrate the generative flexibility of our unconditionally trained model in a wide range of unconditional and conditional generation tasks on benchmark TPPs.

cs.LG↗

arXiv2026

3D Molecule Generation from Rigid Motifs via SE(3) Flows

Three-dimensional molecular structure generation is typically performed at the level of individual atoms, yet molecular graph generation techniques often consider fragments as their structural units. Building on the advances in frame-based protein structure generation, we extend these fragmentation ideas to 3D, treating general molecules as sets of rigid-body motifs. Utilising this representation, we employ SE(3)-equivariant generative modelling for de novo 3D molecule generation from rigid motifs. In our evaluations, we observe comparable or superior results to state-of-the-art across benchmarks, surpassing it in atom stability on GEOM-Drugs, while yielding a 2x to 10x reduction in generation steps and offering 3.5x compression in molecular representations compared to the standard atom-based methods.

cs.LG↗

arXiv2025

GeoDiffusion: A Training-Free Framework for Accurate 3D Geometric Conditioning in Image Generation

Precise geometric control in image generation is essential for engineering \& product design and creative industries to control 3D object features accurately in image space. Traditional 3D editing approaches are time-consuming and demand specialized skills, while current image-based generative methods lack accuracy in geometric conditioning. To address these challenges, we propose GeoDiffusion, a training-free framework for accurate and efficient geometric conditioning of 3D features in image generation. GeoDiffusion employs a class-specific 3D object as a geometric prior to define keypoints and parametric correlations in 3D space. We ensure viewpoint consistency through a rendered image of a reference 3D object, followed by style transfer to meet user-defined appearance specifications. At the core of our framework is GeoDrag, improving accuracy and speed of drag-based image editing on geometry guidance tasks and general instructions on DragBench. Our results demonstrate that GeoDiffusion enables precise geometric modifications across various iterative design workflows.

cs.CV↗

arXiv2025

Efficient Time Series Processing for Transformers and State-Space Models through Token Merging

Despite recent advances in subquadratic attention mechanisms or state-space models, processing long token sequences still imposes significant computational requirements. Token merging has emerged as a solution to increase computational efficiency in computer vision architectures. In this work, we perform the first investigations of token merging in time series analysis on both transformers and state-space models. We further introduce local merging, a domain-specific token merging algorithm that selectively combines tokens within a local neighborhood, achieving two major benefits: a) Local merging can adjust its computational complexity from quadratic to linear based on the neighborhood size to effectively scale to long sequences; b) Local merging is the first causal merging scheme enabling token merging in transformer decoders. Further, we identify spectral properties of the input data that reliably predict the potential benefits of local merging without requiring evaluation on downstream tasks. Our comprehensive empirical evaluation demonstrates that local merging offers substantial efficiency gains with minimal impact on accuracy, achieving up to 5400% acceleration on the recently proposed Chronos foundation model.

cs.LG↗

arXiv2025

Flow Matching with Gaussian Process Priors for Probabilistic Time Series Forecasting

Recent advancements in generative modeling, particularly diffusion models, have opened new directions for time series modeling, achieving state-of-the-art performance in forecasting and synthesis. However, the reliance of diffusion-based models on a simple, fixed prior complicates the generative process since the data and prior distributions differ significantly. We introduce TSFlow, a conditional flow matching (CFM) model for time series combining Gaussian processes, optimal transport paths, and data-dependent prior distributions. By incorporating (conditional) Gaussian processes, TSFlow aligns the prior distribution more closely with the temporal structure of the data, enhancing both unconditional and conditional generation. Furthermore, we propose conditional prior sampling to enable probabilistic forecasting with an unconditionally trained model. In our experimental evaluation on eight real-world datasets, we demonstrate the generative capabilities of TSFlow, producing high-quality unconditional samples. Finally, we show that both conditionally and unconditionally trained models achieve competitive results across multiple forecasting benchmarks.

cs.LG↗

arXiv2025

Assessing Robustness via Score-Based Adversarial Image Generation

Most adversarial attacks and defenses focus on perturbations within small $\ell_p$-norm constraints. However, $\ell_p$ threat models cannot capture all relevant semantics-preserving perturbations, and hence, the scope of robustness evaluations is limited. In this work, we introduce Score-Based Adversarial Generation (ScoreAG), a novel framework that leverages the advancements in score-based generative models to generate unrestricted adversarial examples that overcome the limitations of $\ell_p$-norm constraints. Unlike traditional methods, ScoreAG maintains the core semantics of images while generating adversarial examples, either by transforming existing images or synthesizing new ones entirely from scratch. We further exploit the generative capability of ScoreAG to purify images, empirically enhancing the robustness of classifiers. Our extensive empirical evaluation demonstrates that ScoreAG improves upon the majority of state-of-the-art attacks and defenses across multiple benchmarks. This work highlights the importance of investigating adversarial examples bounded by semantics rather than $\ell_p$-norm constraints. ScoreAG represents an important step towards more encompassing robustness assessments.

cs.CV↗

arXiv2024

Unlocking Point Processes through Point Set Diffusion

Point processes model the distribution of random point sets in mathematical spaces, such as spatial and temporal domains, with applications in fields like seismology, neuroscience, and economics. Existing statistical and machine learning models for point processes are predominantly constrained by their reliance on the characteristic intensity function, introducing an inherent trade-off between efficiency and flexibility. In this paper, we introduce Point Set Diffusion, a diffusion-based latent variable model that can represent arbitrary point processes on general metric spaces without relying on the intensity function. By directly learning to stochastically interpolate between noise and data point sets, our approach enables efficient, parallel sampling and flexible generation for complex conditional tasks defined on the metric space. Experiments on synthetic and real-world datasets demonstrate that Point Set Diffusion achieves state-of-the-art performance in unconditional and conditional generation of spatial and spatiotemporal point processes while providing up to orders of magnitude faster sampling than autoregressive baselines.

cs.LG↗

arXiv2023

Predict, Refine, Synthesize: Self-Guiding Diffusion Models for Probabilistic Time Series Forecasting

Diffusion models have achieved state-of-the-art performance in generative modeling tasks across various domains. Prior works on time series diffusion models have primarily focused on developing conditional models tailored to specific forecasting or imputation tasks. In this work, we explore the potential of task-agnostic, unconditional diffusion models for several time series applications. We propose TSDiff, an unconditionally-trained diffusion model for time series. Our proposed self-guidance mechanism enables conditioning TSDiff for downstream tasks during inference, without requiring auxiliary networks or altering the training procedure. We demonstrate the effectiveness of our method on three different time series tasks: forecasting, refinement, and synthetic data generation. First, we show that TSDiff is competitive with several task-specific conditional forecasting methods (predict). Second, we leverage the learned implicit probability density of TSDiff to iteratively refine the predictions of base forecasters with reduced computational overhead over reverse diffusion (refine). Notably, the generative performance of the model remains intact -- downstream forecasters trained on synthetic samples from TSDiff outperform forecasters that are trained on samples from other state-of-the-art generative time series models, occasionally even outperforming models trained on real data (synthesize).

cs.LG↗

arXiv2022

U-PET: MRI-based Dementia Detection with Joint Generation of Synthetic FDG-PET Images

Alzheimer's disease (AD) is the most common cause of dementia. An early detection is crucial for slowing down the disease and mitigating risks related to the progression. While the combination of MRI and FDG-PET is the best image-based tool for diagnosis, FDG-PET is not always available. The reliable detection of Alzheimer's disease with only MRI could be beneficial, especially in regions where FDG-PET might not be affordable for all patients. To this end, we propose a multi-task method based on U-Net that takes T1-weighted MR images as an input to generate synthetic FDG-PET images and classifies the dementia progression of the patient into cognitive normal (CN), cognitive impairment (MCI), and AD. The attention gates used in both task heads can visualize the most relevant parts of the brain, guiding the examiner and adding interpretability. Results show the successful generation of synthetic FDG-PET images and a performance increase in disease classification over the naive single-task baseline.

eess.IV↗

arXiv2021

Self-Supervised Learning from Unlabeled Fundus Photographs Improves Segmentation of the Retina

Fundus photography is the primary method for retinal imaging and essential for diabetic retinopathy prevention. Automated segmentation of fundus photographs would improve the quality, capacity, and cost-effectiveness of eye care screening programs. However, current segmentation methods are not robust towards the diversity in imaging conditions and pathologies typical for real-world clinical applications. To overcome these limitations, we utilized contrastive self-supervised learning to exploit the large variety of unlabeled fundus images in the publicly available EyePACS dataset. We pre-trained an encoder of a U-Net, which we later fine-tuned on several retinal vessel and lesion segmentation datasets. We demonstrate for the first time that by using contrastive self-supervised learning, the pre-trained network can recognize blood vessels, optic disc, fovea, and various lesions without being provided any labels. Furthermore, when fine-tuned on a downstream blood vessel segmentation task, such pre-trained networks achieve state-of-the-art performance on images from different datasets. Additionally, the pre-training also leads to shorter training times and an improved few-shot performance on both blood vessel and lesion segmentation tasks. Altogether, our results showcase the benefits of contrastive self-supervised pre-training which can play a crucial role in real-world clinical applications requiring robust models able to adapt to new devices with only a few annotated samples.

eess.IV↗

arXiv2021

Geometry-aware neural solver for fast Bayesian calibration of brain tumor models

Modeling of brain tumor dynamics has the potential to advance therapeutic planning. Current modeling approaches resort to numerical solvers that simulate the tumor progression according to a given differential equation. Using highly-efficient numerical solvers, a single forward simulation takes up to a few minutes of compute. At the same time, clinical applications of tumor modeling often imply solving an inverse problem, requiring up to tens of thousands forward model evaluations when used for a Bayesian model personalization via sampling. This results in a total inference time prohibitively expensive for clinical translation. While recent data-driven approaches become capable of emulating physics simulation, they tend to fail in generalizing over the variability of the boundary conditions imposed by the patient-specific anatomy. In this paper, we propose a learnable surrogate for simulating tumor growth which maps the biophysical model parameters directly to simulation outputs, i.e. the local tumor cell densities, whilst respecting patient geometry. We test the neural solver on Bayesian tumor model personalization for a cohort of glioma patients. Bayesian inference using the proposed surrogate yields estimates analogous to those obtained by solving the forward model with a regular numerical solver. The near-real-time computation cost renders the proposed method suitable for clinical settings. The code is available at https://github.com/IvanEz/tumor-surrogate.

cs.CE↗