Sebastián Vollmer

Archana Warrier, Dat Nguyen, Michelangelo Naim et al. · cambridge

Model-learning agents should gather information to learn world models that support many downstream tasks and inferences, such as predicting unobserved states, estimating near- and far-term consequences of actions, planning action sequences, and detecting changes in dynamics. Current methods for learning and evaluating world models diverge from this goal: training and evaluation are anchored to next-frame prediction, and success is scored by reward maximization in the same environment. We propose WorldTest, a protocol to evaluate model-learning agents that separates reward-free interaction from a scored test phase in a different but related environment. WorldTest is open-ended$\unicode{x2014}$models should support many different tasks unknown ahead of time$\unicode{x2014}$and agnostic to model representation, allowing comparison across approaches. We instantiated WorldTest with AutumnBench, a suite of 43 interactive grid-world environments and 129 tasks across three families: masked-frame prediction, planning, and predicting changes to the causal dynamics. We compared 517 human participants and three frontier models on AutumnBench. We found that humans outperform the models, and scaling compute improves performance only in some environments but not others. WorldTest provides a novel template$\unicode{x2014}$reward-free exploration, derived tests, and behavior-based scoring$\unicode{x2014}$to evaluate what agents learn about environment dynamics, and AutumnBench exposes significant headroom in world-model learning.

Santo M. A. R. Thies, Viktor Bengs, Timo Kaufmann et al.

Calibration, the alignment of predicted probabilities with true outcome frequencies, is essential for reliable decision-making. While extensively studied for classification and regression, calibration has not been formally addressed for probabilistic label ranking, where the goal is to predict a distribution over orderings of a label set. Naively treating rankings as classes ignores their structure and fails to capture important modalities such as pairwise and top-k predictions. We formalize calibration for label ranking and develop a hierarchy of notions covering full rankings, sub-rankings, and top-k rankings. We prove that full-rank calibration implies the others but not conversely, and sub-ranking and top-k calibration are incomparable. Empirically, we find popular label ranking models are often poorly calibrated, with substantial differences between sub-ranking and top-k metrics. Applying our framework to RLHF reward models, we find that calibration correlates strongly but not perfectly with benchmark accuracy, suggesting it captures a meaningful quality dimension beyond top-1 accuracy. These findings motivate future work on understanding the downstream effects of miscalibration and developing methods to correct it.

Raphael Sonabend, Andreas Bender, Sebastian Vollmer

In this paper we consider how to evaluate survival distribution predictions with measures of discrimination. This is a non-trivial problem as discrimination measures are the most commonly used in survival analysis and yet there is no clear method to derive a risk prediction from a distribution prediction. We survey methods proposed in literature and software and consider their respective advantages and disadvantages. Whilst distributions are frequently evaluated by discrimination measures, we find that the method for doing so is rarely described in the literature and often leads to unfair comparisons. We find that the most robust method of reducing a distribution to a risk is to sum over the predicted cumulative hazard. We recommend that machine learning survival analysis software implements clear transformations between distribution and risk predictions in order to allow more transparent and accessible model evaluation. The code used in the final experiment is available at https://github.com/RaphaelS1/distribution_discrimination.

Saif Ur Rehman Khan, Muhammad Nabeel Asim, Sebastian Vollmer et al.

The framework is designed to improve performance in the analysis of combined as well as single anatomical perspectives for MRI disease diagnosis. It specifically addresses the performance degradation observed in state-of-the-art (SOTA) models, particularly when processing axial, coronal, and sagittal anatomical planes. The paper introduces the FOLC-Net framework, which incorporates a novel federated-optimized lightweight architecture with approximately 1.217 million parameters and a storage requirement of only 0.9 MB. FOLC-Net integrates Manta-ray foraging optimization (MRFO) mechanisms for efficient model structure generation, global model cloning for scalable training, and ConvNeXt for enhanced client adaptability. The model was evaluated on combined multi-view data as well as individual views, such as axial, coronal, and sagittal, to assess its robustness in various medical imaging scenarios. Moreover, FOLC-Net tests a ShallowFed model on different data to evaluate its ability to generalize beyond the training dataset. The results show that FOLC-Net outperforms existing models, particularly in the challenging sagittal view. For instance, FOLC-Net achieved an accuracy of 92.44% on the sagittal view, significantly higher than the 88.37% accuracy of study method (DL + Residual Learning) and 88.95% of DL models. Additionally, FOLC-Net demonstrated improved accuracy across all individual views, providing a more reliable and robust solution for medical image analysis in decentralized environments. FOLC-Net addresses the limitations of existing SOTA models by providing a framework that ensures better adaptability to individual views while maintaining strong performance in multi-view settings. The incorporation of MRFO, global model cloning, and ConvNeXt ensures that FOLC-Net performs better in real-world medical applications.

Saif Ur Rehman Khan, Muhammad Nabeel Asim, Sebastian Vollmer et al.

Medical disease prediction, particularly through imaging, remains a challenging task due to the complexity and variability of medical data, including noise, ambiguity, and differing image quality. Recent deep learning models, including Knowledge Distillation (KD) methods, have shown promising results in brain tumor image identification but still face limitations in handling uncertainty and generalizing across diverse medical conditions. Traditional KD methods often rely on a context-unaware temperature parameter to soften teacher model predictions, which does not adapt effectively to varying uncertainty levels present in medical images. To address this issue, we propose a novel framework that integrates Ant Colony Optimization (ACO) for optimal teacher-student model selection and a novel context-aware predictor approach for temperature scaling. The proposed context-aware framework adjusts the temperature based on factors such as image quality, disease complexity, and teacher model confidence, allowing for more robust knowledge transfer. Additionally, ACO efficiently selects the most appropriate teacher-student model pair from a set of pre-trained models, outperforming current optimization methods by exploring a broader solution space and better handling complex, non-linear relationships within the data. The proposed framework is evaluated using three publicly available benchmark datasets, each corresponding to a distinct medical imaging task. The results demonstrate that the proposed framework significantly outperforms current state-of-the-art methods, achieving top accuracy rates: 98.01% on the MRI brain tumor (Kaggle) dataset, 92.81% on the Figshare MRI dataset, and 96.20% on the GastroNet dataset. This enhanced performance is further evidenced by the improved results, surpassing existing benchmarks of 97.24% (Kaggle), 91.43% (Figshare), and 95.00% (GastroNet).

Saif Ur Rehman Khan, Muhammad Nabeel Asim, Sebastian Vollmer et al.

This paper presents the FuzzyDistillViT-MobileNet model, a novel approach for lung cancer (LC) classification, leveraging dynamic fuzzy logic-driven knowledge distillation (KD) to address uncertainty and complexity in disease diagnosis. Unlike traditional models that rely on static KD with fixed weights, our method dynamically adjusts the distillation weight using fuzzy logic, enabling the student model to focus on high-confidence regions while reducing attention to ambiguous areas. This dynamic adjustment improves the model ability to handle varying uncertainty levels across different regions of LC images. We employ the Vision Transformer (ViT-B32) as the instructor model, which effectively transfers knowledge to the student model, MobileNet, enhancing the student generalization capabilities. The training process is further optimized using a dynamic wait adjustment mechanism that adapts the training procedure for improved convergence and performance. To enhance image quality, we introduce pixel-level image fusion improvement techniques such as Gamma correction and Histogram Equalization. The processed images (Pix1 and Pix2) are fused using a wavelet-based fusion method to improve image resolution and feature preservation. This fusion method uses the wavedec2 function to standardize images to a 224x224 resolution, decompose them into multi-scale frequency components, and recursively average coefficients at each level for better feature representation. To address computational efficiency, Genetic Algorithm (GA) is used to select the most suitable pre-trained student model from a pool of 12 candidates, balancing model performance with computational cost. The model is evaluated on two datasets, including LC25000 histopathological images (99.16% accuracy) and IQOTH/NCCD CT-scan images (99.54% accuracy), demonstrating robustness across different imaging domains.

Sary Elmansoury, Islam Mesabah, Gerrit Großmann et al.

Vision language models (VLMs) excel at zero-shot visual classification, but their performance on fine-grained tasks and large hierarchical label spaces is understudied. This paper investigates whether structured, tree-based reasoning can enhance VLM performance. We introduce a framework that decomposes classification into interpretable decisions using decision trees and evaluates it on fine-grained (GTSRB) and coarse-grained (CIFAR-10) datasets. Although the model achieves 98.2% accuracy in understanding the tree knowledge, tree-based reasoning consistently underperforms standard zero-shot prompting. We also explore enhancing the tree prompts with LLM-generated classes and image descriptions to improve alignment. The added description enhances the performance of the tree-based and zero-shot methods. Our findings highlight limitations of structured reasoning in visual classification and offer insights for designing more interpretable VLM systems.

Shota Arai, David Selby, Andrew Vargo et al.

Imagine a world where clinical trials need far fewer patients to achieve the same statistical power, thanks to the knowledge encoded in large language models (LLMs). We present a novel framework for hierarchical Bayesian modeling of adverse events in multi-center clinical trials, leveraging LLM-informed prior distributions. Unlike data augmentation approaches that generate synthetic data points, our methodology directly obtains parametric priors from the model. Our approach systematically elicits informative priors for hyperparameters in hierarchical Bayesian models using a pre-trained LLM, enabling the incorporation of external clinical expertise directly into Bayesian safety modeling. Through comprehensive temperature sensitivity analysis and rigorous cross-validation on real-world clinical trial data, we demonstrate that LLM-derived priors consistently improve predictive performance compared to traditional meta-analytical approaches. This methodology paves the way for more efficient and expert-informed clinical trial design, enabling substantial reductions in the number of patients required to achieve robust safety assessment and with the potential to transform drug safety monitoring and regulatory decision making.

Siyi Sun, David Antony Selby, Yunchuan Huang et al.

Missing data imputation in tabular datasets remains a pivotal challenge in data science and machine learning, particularly within socioeconomic research. However, real-world socioeconomic datasets are typically subject to strict data protection protocols, which often prohibit public sharing, even for synthetic derivatives. This severely limits the reproducibility and accessibility of benchmark studies in such settings. Further, there are very few publicly available synthetic datasets. Thus, there is limited availability of benchmarks for systematic evaluation of imputation methods on socioeconomic datasets, whether real or synthetic. In this study, we utilize the World Bank's publicly available synthetic dataset, Synthetic Data for an Imaginary Country, which closely mimics a real World Bank household survey while being fully public, enabling broad access for methodological research. With this as a starting point, we derived the IMAGIC-500 dataset: we select a subset of 500k individuals across approximately 100k households with 19 socioeconomic features, designed to reflect the hierarchical structure of real-world household surveys. This paper introduces a comprehensive missing data imputation benchmark on IMAGIC-500 under various missing mechanisms (MCAR, MAR, MNAR) and missingness ratios (10\%, 20\%, 30\%, 40\%, 50\%). Our evaluation considers the imputation accuracy for continuous and categorical variables, computational efficiency, and impact on downstream predictive tasks, such as estimating educational attainment at the individual level. The results highlight the strengths and weaknesses of statistical, traditional machine learning, and deep learning imputation techniques, including recent diffusion-based methods. The IMAGIC-500 dataset and benchmark aim to facilitate the development of robust imputation algorithms and foster reproducible social science research.

Michael B. Giles, Mateusz B. Majka, Lukasz Szpruch et al.

We develop a framework that allows the use of the multi-level Monte Carlo (MLMC) methodology (Giles2015) to calculate expectations with respect to the invariant measure of an ergodic SDE. In that context, we study the (over-damped) Langevin equations with a strongly concave potential. We show that, when appropriate contracting couplings for the numerical integrators are available, one can obtain a uniform in time estimate of the MLMC variance in contrast to the majority of the results in the MLMC literature. As a consequence, a root mean square error of $\mathcal{O}(\varepsilon)$ is achieved with $\mathcal{O}(\varepsilon^{-2})$ complexity on par with Markov Chain Monte Carlo (MCMC) methods, which however can be computationally intensive when applied to large data sets. Finally, we present a multi-level version of the recently introduced Stochastic Gradient Langevin Dynamics (SGLD) method (Welling and Teh, 2011) built for large datasets applications. We show that this is the first stochastic gradient MCMC method with complexity $\mathcal{O}(\varepsilon^{-2}|\log {\varepsilon}|^{3})$, in contrast to the complexity $\mathcal{O}(\varepsilon^{-3})$ of currently available methods. Numerical experiments confirm our theoretical findings.