arXiv subjects

Göran Kauermann

Publications and source records attributed to Göran Kauermann.

At least 19 recordsLinked to original sources

arXiv2026

Deep Evidential Regression for Sparse Forest Height Estimation from Multimodal Satellite Imagery

Accurate estimation of forest height from satellite imagery is essential for applications such as carbon accounting, biodiversity monitoring, and ecosystem management. While recent deep learning approaches provide accurate predictions, they typically do not quantify predictive uncertainty. This limitation is particularly relevant in geospatial settings characterized by sparse supervision and geographic distribution shift. In this work, we investigate Deep Evidential Regression (DER) for forest height estimation on the TreeUQ benchmark, a large-scale dataset designed for the joint estimation of tree count and average tree height at 10 m resolution, based on Sentinel-1/-2 data as well as tree inventory data over the federal state of Bavaria. To account for the extreme label sparsity of the tree inventory data, we introduce a masked evidential loss for dense geospatial prediction. Using a U-Net architecture with multimodal Sentinel-1 and Sentinel-2 inputs, the proposed approach jointly predicts tree height and associated uncertainty estimates in a single forward pass. Experimental results show that DER achieves predictive performance comparable to a deterministic U-Net while additionally providing well-calibrated uncertainty estimates. These findings demonstrate the potential of evidential learning as an efficient framework for uncertainty-aware forest structure estimation from Earth observation data.

cs.CV↗

arXiv2026

Maximum Likelihood Estimation for Network Models with Latent Geometry under Snowball Sampling

Snowball sampling is a widely used design for collecting network data from large or hard-to-reach populations, yet naive inference that ignores the sampling mechanism produces systematically biased parameter estimates. We derive the exact likelihood of a multi-wave snowball sample for the class of continuous latent space (CLS) models, in which edges form independently conditional on latent vertex-level quantities, and show that conditional edge independence reduces the marginalization over unobserved network configurations to a closed-form expression portable across the entire CLS class. We develop a stochastic Expectation-Maximization algorithm for the Euclidean latent distance model as a concrete implementation, and apply the framework to the large-scale co-inventor network of German semiconductor patent applicants by drawing multiple snowball samples. We find that the naive procedure severely underestimates latent space variance, produces networks with nearly twice the observed edge count, and achieves a spectral goodness-of-fit nine times worse than the corrected model, which directly affects the quantitative interpretation of covariate effects.

stat.ME↗

arXiv2026

Probabilistic Deep Learning for Drought Forecasting: Role of Internal Climate Variability

Predicting drought risk is essential for anticipating impacts on water resources, agriculture, ecosystems, and climate adaptation planning. Yet drought forecasts remain uncertain because variability can substantially alter regional precipitation and evaporative demand. Treating this variability as unstructured noise ignores the fact that internal variability has spatial, seasonal, and temporal structure and thus contains information that can be used to improve drought forecasting. We propose a deep-learning-based forecasting framework for European drought prediction and extend it with an uncertainty-aware drought bound that explicitly incorporates internal forecast variability from a large climate model ensemble. This bound represents a physically plausible lower-tail trajectory of future drought conditions and marks how severe drought could plausibly become under an unfavourable realisation of internal variability, giving adaptation planning a conservative, risk-averse reference. We compare the proposed bound with a lower bound derived from reanalysis data only and show that our proposed ensemble-informed bound is better calibrated across most regions and seasons. This is specifically true during anomalously dry conditions, when historical reanalysis alone underestimates lower-tail drought risk. Our results show that internal variability should be treated as a forecast quantity in its own right. More broadly, large ensembles provide a practical way to transfer physically plausible climate variability into machine-learning drought forecasts, yielding risk-aware bounds that are more informative for drought assessment under shifting climate conditions.

stat.AP↗

arXiv2026

Capturing Aleatoric Uncertainty in Climate Models

Internal climate variability arises from the climate system's inherently chaotic dynamics. Quantifying it is essential for climate science, as it enables risk-based decision-making and differentiates between externally forced change and internal fluctuations. In statistical terms, natural variability corresponds to aleatoric uncertainty, i.e., irreducible stochastic variability. Despite this close conceptual alignment, the link between internal climate variability and aleatoric uncertainty has not yet been formalized. We establish a theoretical link by showing that member-to-member differences in single-model large ensembles provide a direct representation of aleatoric uncertainty. To quantify the spatio-temporal structure of aleatoric uncertainty, we employ generalized additive models. The proposed framework is validated through comparison with ERA5-Land reanalysis data, demonstrating that ensemble-derived estimates reproduce key spatial and temporal patterns of real-world variability. Applied to the water balance over the Iberian Peninsula, our approach reveals coherent variability structures and pronounced regional heterogeneity. We find a decline in variability in drought-prone regions and seasons, a pattern that strengthens under +3 °C global warming, implying an increased risk of persistent summer drought conditions. Beyond this application, the framework is climate-model agnostic and transferable to other variables and spatial scales, providing a statistical basis for quantifying internal climate variability as aleatoric uncertainty.

stat.AP↗

arXiv2026

Assessing Reporting Delays in ACLED Conflict Event Data

Timely and accurate conflict event data are essential for real-time monitoring, forecasting, and policy response. Yet near-real-time conflict datasets such as the Armed Conflict Location \& Event Data Project (ACLED) are subject to reporting delays, that is, delays between event occurrence and first inclusion in the database. Such delays can introduce bias in short-term analyses and forecasts. This study provides a statistical analysis of reporting delays for African events recorded in ACLED's weekly releases from June 30, 2024, to June 1, 2025. Treating delay as a discrete time duration, we estimate grouped proportional hazards models with additive-linear and smooth terms incorporating event-level, spatial, and country-level covariates. Our results show that more than half of events are reported within two weeks, but delays vary systematically by event type, fatalities, geographic location, and political regime. Higher-fatality events are reported more quickly, while events in more restrictive political and informational environments tend to be reported more slowly. We also find substantial between-country heterogeneity, and country-specific analyses indicate that event-level effects differ across contexts. These findings show that reporting delays are structured rather than random and that real-time conflict analysis must account for them. More broadly, they provide an empirical foundation for developing nowcasting approaches to correct short-term underreporting in conflict event data.

stat.AP↗

arXiv2025

Deriving Duration Time from Occupancy Data -- A case study in the length of stay in Intensive Care Units for COVID-19 patients

This paper focuses on drawing information on underlying processes, which are not directly observed in the data. In particular, we work with data in which only the total count of units in a system at a given time point is observed, but the underlying process of inflows, length of stay and outflows is not. The particular data example looked at in this paper is the occupancy of intensive care units (ICU) during the COVID-19 pandemic, where the aggregated numbers of occupied beds in ICUs on the district level (`Landkreis') are recorded, but not the number of incoming and outgoing patients. The Skellam distribution allows us to infer the number of incoming and outgoing patients from the occupancy in the ICUs. This paper goes a step beyond and approaches the question of whether we can also estimate the average length of stay of ICU patients. Hence, the task is to derive not only the number of incoming and outgoing units from a total net count but also to gain information on the duration time of patients on ICUs. We make use of a stochastic Expectation-Maximisation algorithm and additionally include exogenous information which are assumed to explain the intensity of inflow.

stat.AP↗

arXiv2025

Revisiting Active Learning under (Human) Label Variation

Access to high-quality labeled data remains a limiting factor in applied supervised learning. While label variation (LV), i.e., differing labels for the same instance, is common, especially in natural language processing, annotation frameworks often still rest on the assumption of a single ground truth. This overlooks human label variation (HLV), the occurrence of plausible differences in annotations, as an informative signal. Similarly, active learning (AL), a popular approach to optimizing the use of limited annotation budgets in training ML models, often relies on at least one of several simplifying assumptions, which rarely hold in practice when acknowledging HLV. In this paper, we examine foundational assumptions about truth and label nature, highlighting the need to decompose observed LV into signal (e.g., HLV) and noise (e.g., annotation error). We survey how the AL and (H)LV communities have addressed -- or neglected -- these distinctions and propose a conceptual framework for incorporating HLV throughout the AL loop, including instance selection, annotator choice, and label representation. We further discuss the integration of large language models (LLM) as annotators. Our work aims to lay a conceptual foundation for HLV-aware active learning, better reflecting the complexities of real-world annotation.

cs.CL↗

arXiv2025

Structure Is Not Enough: Leveraging Behavior for Neural Network Weight Reconstruction

The weights of neural networks (NNs) have recently gained prominence as a new data modality in machine learning, with applications ranging from accuracy and hyperparameter prediction to representation learning or weight generation. One approach to leverage NN weights involves training autoencoders (AEs), using contrastive and reconstruction losses. This allows such models to be applied to a wide variety of downstream tasks, and they demonstrate strong predictive performance and low reconstruction error. However, despite the low reconstruction error, these AEs reconstruct NN models with deteriorated performance compared to the original ones, limiting their usability with regard to model weight generation. In this paper, we identify a limitation of weight-space AEs, specifically highlighting that a structural loss, that uses the Euclidean distance between original and reconstructed weights, fails to capture some features critical for reconstructing high-performing models. We analyze the addition of a behavioral loss for training AEs in weight space, where we compare the output of the reconstructed model with that of the original one, given some common input. We show a strong synergy between structural and behavioral signals, leading to increased performance in all downstream tasks evaluated, in particular NN weights reconstruction and generation.

cs.LG↗

arXiv2025

Sources of Uncertainty in Supervised Machine Learning -- A Statisticians' View

Supervised machine learning and predictive models have achieved an impressive standard today, enabling us to answer questions that were inconceivable a few years ago. Besides these successes, it becomes clear, that beyond pure prediction, which is the primary strength of most supervised machine learning algorithms, the quantification of uncertainty is relevant and necessary as well. However, before quantification is possible, types and sources of uncertainty need to be defined precisely. While first concepts and ideas in this direction have emerged in recent years, this paper adopts a conceptual, basic science perspective and examines possible sources of uncertainty. By adopting the viewpoint of a statistician, we discuss the concepts of aleatoric and epistemic uncertainty, which are more commonly associated with machine learning. The paper aims to formalize the two types of uncertainty and demonstrates that sources of uncertainty are miscellaneous and can not always be decomposed into aleatoric and epistemic. Drawing parallels between statistical concepts and uncertainty in machine learning, we emphasise the role of data and their influence on uncertainty.

stat.ML↗

arXiv2024

Using LASSO for Variable Selection in Exponential Random Graph models

The paper demonstrates the use of LASSO-based estimation in network models. Taking the Exponential Random Graph Model (ERGM) as a flexible and widely used model for network data analysis, the paper focuses on the question of how to specify the (sufficient) statistics, that define the model structure. This includes both, endogenous network statistics (e.g. twostars, triangles, etc.) as well as statistics involving exogenous covariates; on the node as well as on the edge level. LASSO estimation is a penalized estimation that shrinks some of the parameter estimates to be equal to zero. As such it allows for model selection by modifying the amount of penalty. The concept is well established in standard regression and we demonstrate its usage in network data analysis, with the advantage of automatically providing a model selection framework.

stat.ME↗

arXiv2024

Human-in-the-loop: Towards Label Embeddings for Measuring Classification Difficulty

Uncertainty in machine learning models is a timely and vast field of research. In supervised learning, uncertainty can already occur in the first stage of the training process, the annotation phase. This scenario is particularly evident when some instances cannot be definitively classified. In other words, there is inevitable ambiguity in the annotation step and hence, not necessarily a "ground truth" associated with each instance. The main idea of this work is to drop the assumption of a ground truth label and instead embed the annotations into a multidimensional space. This embedding is derived from the empirical distribution of annotations in a Bayesian setup, modeled via a Dirichlet-Multinomial framework. We estimate the model parameters and posteriors using a stochastic Expectation Maximization algorithm with Markov Chain Monte Carlo steps. The methods developed in this paper readily extend to various situations where multiple annotators independently label instances. To showcase the generality of the proposed approach, we apply our approach to three benchmark datasets for image classification and Natural Language Inference. Besides the embeddings, we can investigate the resulting correlation matrices, which reflect the semantic similarities of the original classes very well for all three exemplary datasets.

cs.LG↗

arXiv2024

Predictive Energy Management for Battery Electric Vehicles with Hybrid Models

This paper addresses the problem of predicting the energy consumption for the drivers of Battery electric vehicles (BEVs). Several external factors (e.g., weather) are shown to have huge impacts on the energy consumption of a vehicle besides the vehicle or powertrain dynamics. Thus, it is challenging to take all of those influencing variables into consideration. The proposed approach is based on a hybrid model which improves the prediction accuracy of energy consumption of BEVs. The novelty of this approach is to combine a physics-based simulation model, which captures the basic vehicle and powertrain dynamics, with a data-driven model. The latter accounts for other external influencing factors neglected by the physical simulation model, using machine learning techniques, such as generalized additive mixed models, random forests and boosting. The hybrid modeling method is evaluated with a real data set from TUM and the hybrid models were shown that decrease the average prediction error from 40% of the pure physics model to 10%.

eess.SY↗

arXiv2023

Estimating excess mortality in high-income countries during the COVID-19 pandemic

Quantifying the number of deaths caused by the COVID-19 crisis has been an ongoing challenge for scientists, and no golden standard to do so has yet been established. We propose a principled approach to calculate age-adjusted yearly excess mortality, and apply it to obtain estimates and uncertainty bounds for 30 countries with publicly available data. The results uncover remarkable variation in pandemic outcomes across different countries. We further compare our findings with existing estimates published in other major scientific outlets, highlighting the importance of proper age adjustment to obtain unbiased figures.

stat.AP↗

arXiv2023

Actor Heterogeneity and Explained Variance in Network Models -- A Scalable Approach through Variational Approximations

The analysis of network data has gained considerable interest in recent years. This also includes the analysis of large, high-dimensional networks with hundreds and thousands of nodes. While exponential random graph models serve as workhorse for network data analyses, their applicability to very large networks is problematic via classical inference such as maximum likelihood or exact Bayesian estimation owing to scaling and instability issues. The latter trace from the fact that classical network statistics consider nodes as exchangeable, i.e., actors in the network are assumed to be homogeneous. This is often questionable. One way to circumvent the restrictive assumption is to include actor-specific random effects, which account for unobservable heterogeneity. However, this increases the number of unknowns considerably, thus making the model highly-parameterized. As a solution even for very large networks, we propose a scalable approach based on variational approximations, which not only leads to numerically stable estimation but is also applicable to high-dimensional directed as well as undirected networks. We furthermore demonstrate that including node-specific covariates can reduce node heterogeneity, which we facilitate through versatile prior formulations and a new measure that we call posterior explained variance. We illustrate our approach in three diverse examples, covering network data from the Italian Parliament, international arms trading, and Facebook; and conduct detailed simulation studies.

stat.ME↗

arXiv2023

Categorising the World into Local Climate Zones -- Towards Quantifying Labelling Uncertainty for Machine Learning Models

Image classification is often prone to labelling uncertainty. To generate suitable training data, images are labelled according to evaluations of human experts. This can result in ambiguities, which will affect subsequent models. In this work, we aim to model the labelling uncertainty in the context of remote sensing and the classification of satellite images. We construct a multinomial mixture model given the evaluations of multiple experts. This is based on the assumption that there is no ambiguity of the image class, but apparently in the experts' opinion about it. The model parameters can be estimated by a stochastic Expectation Maximization algorithm. Analysing the estimates gives insights into sources of label uncertainty. Here, we focus on the general class ambiguity, the heterogeneity of experts, and the origin city of the images. The results are relevant for all machine learning applications where image classification is pursued and labelling is subject to humans.

stat.AP↗

arXiv2023

COVID-19 and social media: Beyond polarization

The COVID-19 pandemic brought upon a massive wave of disinformation, exacerbating polarization in the increasingly divided landscape of online discourse. In this context, popular social media users play a major role, as they have the ability to broadcast messages to large audiences and influence public opinion. In this paper, we make use of openly available data to study the behavior of popular users discussing the pandemic on Twitter. We tackle the issue from a network perspective, considering users as nodes and following relationships as directed edges. The resulting network structure is modeled by embedding the actors in a latent social space, where users closer to one another have a higher probability of following each other. The results suggest the existence of two distinct communities, which can be interpreted as "generally pro" and "generally against" vaccine mandates, corroborating existing evidence on the pervasiveness of echo chambers on the platform. By focusing on a number of notable users, such as politicians, activists, and news outlets, we further show that the two groups are not entirely homogeneous, and that not just the two poles are represented. To the contrary, the latent space captures an entire spectrum of beliefs between the two extremes, demonstrating that polarization, while present, is not the only driver of the network, and that more moderate, "central" users are key players in the discussion.

cs.SI↗

arXiv2023

Dependence matters: Statistical models to identify the drivers of tie formation in economic networks

Networks are ubiquitous in economic research on organizations, trade, and many other areas. However, while economic theory extensively considers networks, no general framework for their empirical modeling has yet emerged. We thus introduce two different statistical models for this purpose -- the Exponential Random Graph Model (ERGM) and the Additive and Multiplicative Effects network model (AME). Both model classes can account for network interdependencies between observations, but differ in how they do so. The ERGM allows one to explicitly specify and test the influence of particular network structures, making it a natural choice if one is substantively interested in estimating endogenous network effects. In contrast, AME captures these effects by introducing actor-specific latent variables affecting their propensity to form ties. This makes the latter a good choice if the researcher is interested in capturing the effect of exogenous covariates on tie formation without having a specific theory on the endogenous dependence structures at play. After introducing the two model classes, we showcase them through real-world applications to networks stemming from international arms trade and foreign exchange activity. We further provide full replication materials to facilitate the adoption of these methods in empirical economic research.

stat.ME↗

arXiv2023

The Politics of Language Choice: How the Russian-Ukrainian War Influences Ukrainians' Language Use on Twitter

The use of language is innately political and often a vehicle of cultural identity as well as the basis for nation building. Here, we examine language choice and tweeting activity of Ukrainian citizens based on more than 4 million geo-tagged tweets from over 62,000 users before and during the Russian-Ukrainian War, from January 2020 to October 2022. Using statistical models, we disentangle sample effects, arising from the in- and outflux of users on Twitter, from behavioural effects, arising from behavioural changes of the users. We observe a steady shift from the Russian language towards the Ukrainian language already before the war, which drastically speeds up with its outbreak. We attribute these shifts in large part to users' behavioural changes. Notably, we find that more than half of the Russian-tweeting users shift towards Ukrainian as a result of the war.

cs.CY↗