14 papers
Walking the Score Manifold: Continuous-time Generative Dynamics on Learned Data Manifolds
Jan Tauberschmidt, Brian B. Moser, Stanislav Frolov +3
Generative modeling of time-dependent data is typically formulated on a discrete temporal grid, restricting supervision to the observed timestamps in the training data. We instead…
Soft Specialists: -Rényi Ensembles for Uncertainty-Aware LLM Post-Training
Paula Cordero-Encinar, Georgy Tyukin, Andrew B. Duncan
Existing training approaches for large language models learn a single set of parameters, based on large volumes of data, which is typically heterogeneous, conflicting and often out…
Batch-based Bayesian Optimal Experimental Design in Linear Inverse Problems
Sofia Mäkinen, Andrew B. Duncan, Tapio Helin
Experimental design is central to science and engineering. A ubiquitous challenge is how to maximize the value of information obtained from expensive or constrained experimental se…
Diffusion Path Samplers via Sequential Monte Carlo
James Matthew Young, Paula Cordero-Encinar, Sebastian Reich +2
We develop diffusion-based samplers for target distributions known up to a normalising constant. To this end, we rely on the well-known diffusion path that smoothly interpolates be…
Learning Density Functionals to Bridge Particle and Continuum Scales
Edoardo Monti, Peter Yatsyshin, Konstantinos Gkagkas +1
Predicting interfacial thermodynamics across molecular and continuum scales remains a central challenge in computational science. Classical density functional theory (cDFT) provide…
How to Train Private Clinical Language Models: A Comparative Study of Privacy-Preserving Pipelines for ICD-9 Coding
Mathieu Dufour, Andrew Duncan
Large language models trained on clinical text risk exposing sensitive patient information, yet differential privacy (DP) methods often severely degrade the diagnostic accuracy nee…
Certified Self-Consistency: Statistical Guarantees and Test-Time Training for Reliable Reasoning in LLMs
Paula Cordero-Encinar, Andrew B. Duncan
Recent advances such as self-consistency and test-time reinforcement learning (TTRL) improve the reliability of large language models (LLMs) without additional supervision, yet the…
Uniform-in-time convergence bounds for Persistent Contrastive Divergence algorithms
Paul Felix Valsecchi Oliva, O. Deniz Akyildiz, Andrew Duncan
We propose a continuous-time formulation of a noisy persistent contrastive divergence (PCD)-like method for maximum likelihood estimation (MLE) of unnormalised densities. Our appro…
Sampling as Bandits: Evaluation-Efficient Design for Black-Box Densities
Takuo Matsubara, Andrew Duncan, Simon Cotter +1
We propose bandit importance sampling (BIS), a powerful importance sampling framework tailored for settings in which evaluating the target density is computationally expensive. BIS…
Sampling by averaging: A multiscale approach to score estimation
Paula Cordero-Encinar, Andrew B. Duncan, Sebastian Reich +1
We introduce a novel framework for efficient sampling from complex, unnormalised target distributions by exploiting multiscale dynamics. Traditional score-based sampling methods ei…
Physics-Constrained Fine-Tuning of Flow-Matching Models for Generation and Inverse Problems
Jan Tauberschmidt, Sophie Fellenz, Sebastian J. Vollmer +1
We present a framework for fine-tuning flow-matching generative models to enforce physical constraints and solve inverse problems in scientific systems. Starting from a model train…
Non-asymptotic Analysis of Diffusion Annealed Langevin Monte Carlo for Generative Modelling
Paula Cordero-Encinar, O. Deniz Akyildiz, Andrew B. Duncan
We investigate the theoretical properties of general diffusion (interpolation) paths and their Langevin Monte Carlo implementation, referred to as diffusion annealed Langevin Monte…
Energy-Based Modelling for Discrete and Mixed Data via Heat Equations on Structured Spaces
Tobias Schröder, Zijing Ou, Yingzhen Li +1
Energy-based models (EBMs) offer a flexible framework for probabilistic modelling across various data domains. However, training EBMs on data in discrete or mixed state spaces pose…
Deep Optimal Sensor Placement for Black Box Stochastic Simulations
Paula Cordero-Encinar, Tobias Schröder, Peter Yatsyshin +1
Selecting cost-effective optimal sensor configurations for subsequent inference of parameters in black-box stochastic systems faces significant computational barriers. We propose a…