Autoregressive graph generators define likelihoods via a sequential construction process, but these likelihoods are only meaningful if they are consistent across all linearizations of the same graph. Segmented Eulerian Neighborhood Trails (SENT), a recent linearization method, converts graphs into sequences that can be perfectly decoded and efficiently processed by language models, but admit multiple equivalent linearizations of the same graph. We quantify violations in assigned negative log-likelihood (NLL) using the coefficient of variation across equivalent linearizations, which we call Linearization Uncertainty (LU). Training transformers under four linearization strategies on two datasets, we show that biased orderings achieve lower NLL on their native order but exhibit expected calibration error (ECE) two orders of magnitude higher under random permutation, indicating that these models have learned their training linearization rather than the underlying graph. On the molecular graph benchmark QM9, NLL for generated graphs is negatively correlated with molecular stability (AUC =0.43), while LU achieves AUC =0.85, suggesting that permutation-based evaluation provides a more reliable quality check for generated molecules. Code is available at https://github.com/lauritsf/linearization-uncertainty

We present ZoRRO (Zero-Weight Personalized Recommender System), a zero-weight, training-free framework for personalized news recommendation designed for scalable real-world deployment. ZoRRO outperforms strong neural baselines in offline ranking evaluations and achieves click-through rate performance in online A/B testing that is nearly on par with a state-of-the-art deep learning model, while operating more than 600 times faster. Our experiments reveal gaps between offline and online performance and demonstrate that models with similar click-through rate outcomes can produce markedly different recommendation distributions, thereby influencing the overall news flow. These findings position ZoRRO as a practical and efficient solution for large-scale news recommendation and highlight the importance of evaluating recommender systems using metrics beyond accuracy alone.

Uncertainty quantification (UQ) in deep learning regression is of wide interest, as it supports critical applications including sequential decision making and risk-sensitive tasks. In heteroskedastic regression, where the uncertainty of the target depends on the input, a common approach is to train a neural network that parameterizes the mean and the variance of the predictive distribution. Still, training deep heteroskedastic regression models poses practical challenges in the trade-off between uncertainty quantification and mean prediction, such as optimization difficulties, representation collapse, and variance overfitting. In this work we identify previously undiscussed fallacies and propose a simple and efficient procedure that addresses these challenges jointly by post-hoc fitting a variance model across the intermediate layers of a pretrained network on a hold-out dataset. We demonstrate that our method achieves on-par or state-of-the-art uncertainty quantification on several molecular graph datasets, without compromising mean prediction accuracy and remaining cheap to use at prediction time.

In the RecSys ’24 Challenge, participants tackled news recommendations using a large-scale Danish dataset. Although the top-performing models achieved similar accuracy scores, they produced markedly different results on beyond-accuracy metrics, highlighting the need for further research into the normative alignment of recommender systems.

Recommender systems optimized solely for user engagement often fail to meet broader normative objectives such as fairness, diversity, or editorial values. We introduce NAILS (Normative Alignment of recommender systems via Internal Label Shift), a simple and scalable method for aligning recommendation outputs with target distributions over item-level attributes (e.g., categories). NAILS modifies the user-conditional item distribution to induce a specified marginal distribution over attributes, leveraging existing user-item preferences without retraining the model. To achieve this, we recast the problem as a form of label shift applied internally within a hierarchical classification framework. Adopting a stakeholdercentric perspective, NAILS enables alignment with global normative goals. Empirically, we show that NAILS consistently improves attribute-level alignment with minimal impact on user engagement, providing a practical mechanism for value-driven recommendation. Our code is available at https://github.com/johanneskruse/nails.

Bayesian Neural Networks (BNNs) often improve model calibration and predictive uncertainty quantification compared to point estimators such as maximum-a-posteriori (MAP). Similarly, deep ensembles (DEs) are also known to improve calibration, and therefore, it is natural to hypothesize that deep ensembles of BNNs (DE-BNNs) should provide even further improvements. In this work, we systematically investigate this across a number of datasets, neural network architectures, and BNN approximation methods and surprisingly find that when the ensembles grow large enough, DEs consistently outperform DE-BNNs on in-distribution data. To shine light on this observation, we conduct several sensitivity and ablation studies. Moreover, we show that even though DE-BNNs outperform DEs on out-of-distribution metrics, this comes at the cost of decreased in-distribution performance. As a final contribution, we open-source the large pool of trained models to facilitate further research on this topic.

Bayesian deep learning provides a framework for quantifying uncertainty. However, the scale of modern neural networks applied in Natural Language Processing (NLP) limits the usability of Bayesian methods. Subnetwork inference aims to approximate the posterior by selecting a stochastic parameter subset for inference, thereby allowing scalable posterior approximations. Determining the optimal parameter space for subnetwork inference is far from trivial. In this paper, we study partially stochastic Bayesian neural networks in the context of transformer models for NLP tasks for the Laplace approximation (LA) and Stochastic weight averaging - Gaussian (SWAG). We propose heuristics for selecting which layers to include in the stochastic subset. We show that norm-based selection is promising for small subsets, and random selection is superior for larger subsets. Moreover, we propose Sparse-KFAC (S-KFAC), an extension of KFAC LA, which selects dense stochastic substructures of linear layers based on parameter magnitudes. S-KFAC retains performance while requiring substantially fewer stochastic parameters and, therefore, drastically limits memory footprint.

Personalized content recommendations have been pivotal to the content experience in digital media from video streaming to social networks. However, several domain specific challenges have held back adoption of recommender systems in news publishing. To address these challenges, we introduce the Ekstra Bladet News Recommendation Dataset (EB-NeRD). The dataset encompasses data from over a million unique users and more than 37 million impression logs from Ekstra Bladet. It also includes a collection of over 125,000 Danish news articles, complete with titles, abstracts, bodies, and metadata, such as categories. EB-NeRD served as the benchmark dataset for the RecSys '24 Challenge, where it was demonstrated how the dataset can be used to address both technical and normative challenges in designing effective and responsible recommender systems for news publishing. The dataset is available at: https://recsys.eb.dk.

In remote sensing there exists a common need for learning scale invariant shapes of objects like buildings. Prior works relies on tweaking multiple loss functions to convert segmentation maps into the final scale invariant representation, necessitating arduous design and optimization. For this purpose we introduce the GeoFormer, a novel architecture which presents a remedy to the said challenges, learning to generate multipolygons end-to-end. By modeling keypoints as spatially dependent tokens in an auto-regressive manner, the GeoFormer outperforms existing works in delineating building objects from satellite imagery. We evaluate the robustness of the GeoFormer against former methods through a variety of parameter ablations and highlight the advantages of optimizing a single likelihood function. Our study presents the first successful application of auto-regressive transformer models for multi-polygon predictions in remote sensing, suggesting a promising methodological alternative for building vectorization.

Black-box variational inference (BBVI) now sees widespread use in machine learning and statistics as a fast yet flexible alternative to Markov chain Monte Carlo methods for approximate Bayesian inference. However, stochastic optimization methods for BBVI remain unreliable and require substantial expertise and hand-tuning to apply effectively. In this paper, we propose robust and automated black-box VI (RABVI), a framework for improving the reliability of BBVI optimization. RABVI is based on rigorously justified automation techniques, includes just a small number of intuitive tuning parameters, and detects inaccurate estimates of the optimal variational approximation. RABVI adaptively decreases the learning rate by detecting convergence of the fixed-learning-rate iterates, then estimates the symmetrized Kullback-Leibler (KL) divergence between the current variational approximation and the optimal one. It also employs a novel optimization termination criterion that enables the user to balance desired accuracy against computational cost by comparing (i) the predicted relative decrease in the symmetrized KL divergence if a smaller learning were used and (ii) the predicted computation required to converge with the smaller learning rate. We validate the robustness and accuracy of RABVI through carefully designed simulation studies and on a diverse set of real-world model and data examples.

We present GAST, a novel model for realistic building delineation, trained using noisy, imperfect ortho imagery and designed for real-life applications. While most popular methods today rely on some form of semantic segmentation, the core interest is not the building's interior points but rather the sequence of points surrounding the outer hull, i.e., the most sparse set of points encapsulating the geometry of the building. Our method works end-to-end, removing the need for post-processing, while demonstrating generalization across large geographical differences. We compare our method to state-of-the-art, complementary works and demonstrate that our model outperforms the baselines in a variety of circumstances and across all metrics relating to polygon fidelity. We release our dataset and model checkpoints at www.huggingface.co/datasets/pihalf/ERBD

The RecSys Challenge 2024 aims to advance news recommendation by addressing both the technical and normative challenges inherent in designing effective and responsible recommender systems for news publishing. This paper describes the challenge, including its objectives, problem setting, and the dataset provided by the Danish news publishers Ekstra Bladet and JP/Politikens Media Group ("Ekstra Bladet"). The challenge explores the unique aspects of news recommendation, such as modeling user preferences based on behavior, accounting for the influence of the news agenda on user interests, and managing the rapid decay of news items. Additionally, the challenge embraces normative complexities, investigating the effects of recommender systems on news flow and their alignment with editorial values. We summarize the challenge setup, dataset characteristics, and evaluation metrics. Finally, we announce the winners and highlight their contributions. The dataset is available at: https://recsys.eb.dk.

Score-based models, trained with denoising score matching, are remarkably effective in generating high dimensional data. However, the high variance of their training objective hinders optimisation. We attempt to reduce it with a control variate, derived via a k-th order Taylor expansion on the training objective and its gradient. We prove an equivalence between the two and demonstrate empirically the effectiveness of our approach on a low dimensional problem setting; and study its effect on larger problems.

In this work, we address the problem of assessing and constructing feedback for early-stage writing automatically using machine learning. Early-stage writing is typically vastly different from conventional writing due to phonetic spelling and lack of proper grammar, punctuation, spacing etc. Consequently, early-stage writing is highly non-trivial to analyze using common linguistic metrics. We propose to use sequence-to-sequence models for "translating" early-stage writing by students into "conventional" writing, which allows the translated text to be analyzed using linguistic metrics. Furthermore, we propose a novel robust likelihood to mitigate the effect of noise in the dataset. We investigate the proposed methods using a set of numerical experiments and demonstrate that the conventional text can be predicted with high accuracy.

Anecdotal evidence suggests that pancreatic acinar metaplasia (PAM) and intestinal metaplasia (IM) overlap infrequently at the gastroesophageal junction/distal esophagus (GEJ/DE). The goal of this study was to evaluate the significance of PAM at GEJ/DE in relation to IM in patients with gastroesophageal reflux disease (GERD). Group 1 comprised 230 consecutive patients with GEJ/DE biopsies (80.6% with GERD symptoms). Group 2 comprised 151 patients with established GERD and GEJ/DE biopsies taken before Nissen fundoplication. Group 3 comprised 540 consecutive patients used for a follow-up study of PAM. PAM was present in 15.7%-15.9% and IM in 24.8%-31.1% of patients in groups 1 and 2, respectively. PAM-IM overlap was present in 2.2%-3.3%, respectively. Patients with PAM were, on average, 6-12 years younger than patients with IM, and were predominantly female (72.2%-75%), in contrast to patients with IM (47.3%-32%). In the unadjusted logistic regression model, patients with PAM were 69%-65% less likely to also have IM, as compared to patients without PAM. In the fully adjusted model, patients with PAM were 35%-61% less likely to also have IM, although the P-value was not significant. Follow-up analysis of patients with PAM from group 3 (n Z 28) demonstrated the prev-alence of IM and PAM in subsequent biopsies at 7.1% and 60.7%, respectively. No cases showed PAM-IM overlap on follow-up. The data suggests that PAM at the GEJ/DE is associated with protective effect against IM and thus could be useful as a marker of decreased susceptibility to IM.& COPY; 2023 Elsevier Inc. All rights reserved.

With the rise of algorithmic personalization, news organizations are finding it necessary to entrust traditionally held editorial values, such as prioritizing news for readers, to automated systems. In a case study conducted by Ekstra Bladet, the Platform Intelligent News project demonstrates how recommender systems successfully improved the click-through rates for various segments on ekstrabladet.dk, while still maintaining the news organization’s editorial values.

We present a novel pipeline for learning the conditional distribution of a building roof mesh given pixels from an aerial image, under the assumption that roof geometry follows a set of regular patterns. Unlike alternative methods that require multiple images of the same object, our approach enables estimating 3D roof meshes using only a single image for predictions. The approach employs the PolyGen, a deep generative transformer architecture for 3D meshes. We apply this model in a new domain and investigate the sensitivity of the image resolution. We propose a novel metric to evaluate the performance of the inferred meshes, and our results show that the model is robust even at lower resolutions, while qualitatively producing realistic representations for out-of-distribution samples.

In this work, we study the feasibility of using machine learning and natural language processing methods for assessing writing proficiency in Danish with respect to text construction, sentence construction, and use of modifiers. Our work is based on the analytical framework for scoring early writing proposed by Kabel et al. (2022), where each text is first annotated by a human expert according to a predefined coding scheme and subsequently scored using statistical Rasch modeling (Rasch, 1960). We investigate two different strategies for estimating these scores automatically: 1) we propose a system for identifying the central linguistic features automatically mimicking the role of the human experts and 2) we train state-of-the-art discriminative machine learning models to predict the proficiency scores directly from the texts. We conduct a number of experiments to evaluate and compare the two approaches. Our results show strong and statistically significant correlations between the scores generated using the automatic system and scores based on human experts. We also estimate and report the reliability of the individual linguistic features in the automatic annotation system. Finally, we also propose and evaluate an extension of the statistical model, which allows the model to compensate for potential systematic errors in the automatic annotations. The article thereby contributes to the area of automated essay scoring (AES) and shows that it is possible to provide teachers with automated valid and reliable knowledge about the development of their students' writing competences, which they can use in their feedback to students.

Bayesian optimization (BO) is a popular method for black-box optimization, which relies on uncertainty as part of its decision-making process when deciding which experiment to perform next. However, not much work has addressed the effect of uncertainty on the performance of the BO algorithm and to what extent calibrated uncertainties improve the ability to find the global optimum. In this work, we provide an extensive study of the relationship between the BO performance (regret) and uncertainty calibration for popular surrogate models and compare them across both synthetic and real-world experiments. Our results confirm that Gaussian Processes are strong surrogate models and that they tend to outperform other popular models. Our results further show a positive association between calibration error and regret, but interestingly, this association disappears when we control for the type of model in the analysis. We also studied the effect of re-calibration and demonstrate that it generally does not lead to improved regret. Finally, we provide theoretical justification for why uncertainty calibration might be difficult to combine with BO due to the small sample sizes commonly used.

In geospatial planning, it is often essential to represent objects in a vectorized format, as this format easily translates to downstream tasks such as web development, graphics, or design. While these problems are frequently addressed using semantic segmentation, which requires additional post-processing to vectorize objects in a non-trivial way, we present an Image-to-Sequence model that allows for direct shape inference and is ready for vector-based workflows out of the box. We demonstrate the model's performance in various ways, including perturbations to the image input that correspond to variations or artifacts commonly encountered in remote sensing applications. Our model outperforms prior works when using ground truth bounding boxes (one object per image), achieving the lowest maximum tangent angle error.