Understanding driver interactions is critical to designing autonomous vehicles to interoperate safely with human-driven cars. We consider the impact of these interactions on the policies drivers employ when navigating unsigned intersections in a driving simulator. The simulator allows the collection of naturalistic decision-making and behavior data in a controlled environment. Using these data, we model the human driver responses as state-based feedback controllers learned via Gaussian Process regression methods. We compute the feedback gain of the controller using a weighted combination of linear and nonlinear priors. We then analyze how the individual gains are reflected in driver behavior. We also assess differences in these controllers across populations of drivers. Our work in data-driven analyses of how drivers determine their policies can facilitate future work in the design of socially responsive autonomy for vehicles.

Deep learning models for medical image segmentation suffer significant performance drops due to distribution shifts, but the causal mechanisms behind these drops remain poorly understood. We extend causal attribution frameworks to high-dimensional segmentation tasks, quantifying how acquisition protocols and annotation variability independently contribute to performance degradation. We model the data-generating process through a causal graph and employ Shapley values to fairly attribute performance changes to individual mechanisms. Our framework addresses unique challenges in medical imaging: high-dimensional outputs, limited samples, and complex mechanism interactions. Validation on multiple sclerosis (MS) lesion segmentation across 4 centers and 7 annotators reveals context-dependent failure modes: annotation protocol shifts dominate when crossing annotators (7.4% $\pm$ 8.9% DSC attribution), while acquisition shifts dominate when crossing imaging centers (6.5% $\pm$ 9.1%). This mechanism-specific quantification enables practitioners to prioritize targeted interventions based on deployment context.

Hallucinations are outputs by Large Language Models (LLMs) that are factually incorrect yet appear plausible [1]. This paper investigates how such hallucinations influence users' trust in LLMs and users' interaction with LLMs. To explore this in everyday use, we conducted a qualitative study with 192 participants. Our findings show that hallucinations do not result in blanket mistrust but instead lead to context-sensitive trust calibration. Building on the calibrated trust model by Lee & See [2] and Afroogh et al.'s trust-related factors [3], we confirm expectancy [3], [4], prior experience [3], [4], [5], and user expertise & domain knowledge [3], [4] as userrelated (human) trust factors, and identify intuition as an additional factor relevant for hallucination detection. Additionally, we found that trust dynamics are further influenced by contextual factors, particularly perceived risk [3] and decision stakes [6]. Consequently, we validate the recursive trust calibration process proposed by Blöbaum [7] and extend it by including intuition as a user-related trust factor. Based on these insights, we propose practical recommendations for responsible and reflective LLM use.

Narratives about artificial intelligence (AI) entangle autonomy, the capacity to self-govern, with sentience, the capacity to sense and feel. AI agents that perform tasks autonomously and companions that recognize and express emotions may activate mental models of autonomy and sentience, respectively, provoking distinct reactions. To examine this possibility, we conducted three pilot studies (N = 374) and four preregistered vignette experiments describing an AI as autonomous, sentient, both, or neither (N = 2,702). Activating a mental model of sentience increased general mind perception (cognition and emotion) and moral consideration more than autonomy, but autonomy increased perceived threat more than sentience. Sentience also increased perceived autonomy more than vice versa. Based on a within-paper meta-analysis, sentience changed reactions more than autonomy on average. By disentangling different mental models of AI, we can study human-AI interaction with more precision to better navigate the detailed design of anthropomorphized AI and prompting interfaces.

Severe heatwaves in urban areas significantly threaten public health, calling for establishing early warning strategies. Despite predicting occurrence of heatwaves and attributing historical mortality, predicting an incoming deadly heatwave remains a challenge due to the difficulty in defining and estimating heat-related mortality. Furthermore, establishing an early warning system imposes additional requirements, including data availability, spatial and temporal robustness, and decision costs. To address these challenges, we propose DeepTherm, a modular early warning system for deadly heatwave prediction without requiring heat-related mortality history. By highlighting the flexibility of deep learning, DeepTherm employs a dual-prediction pipeline, disentangling baseline mortality in the absence of heatwaves and other irregular events from all-cause mortality. We evaluated DeepTherm on real-world data across Spain. Results demonstrate consistent, robust, and accurate performance across diverse regions, time periods, and population groups while allowing trade-off between missed alarms and false alarms.

Neural networks, while effective for tackling many challengi ng scientific tasks, are not known to perform well out-of-distribution (OOD), i.e., within domains which d iffer from their training data. Understanding neural network OOD generalization is paramount to their suc cessful deployment in experimental workflows, especially when ground-truth knowledge about the experime nt is hard to establish or experimental conditions significantly vary. With inherent access to ground-truth in formation and fine-grained control of underlying distributions, simulation-based data curation facilitate s precise investigation of OOD generalization behavior. Here, we probe generalization with respect to imaging condi tions of neural network segmentation models for high-resolution transmission electron microscopy (HRTEM) imaging of nanoparticles, training and measuring the OOD generalization of over 12,000 neural networks using synthetic data generated via random structure sampling and multislice simulation. Using the HRTEM contra st transfer function, we further develop a framework to compare information content of HRTEM datasets an d quantify OOD domain shifts. We demonstrate that neural network segmentation models enjoy significant performance stability, but will smoothly and predictably worsen as imaging conditions shift from the training distribution. Lastly, we consider limitations of our approach in explaining other OOD shifts, s uch as of the atomic structures, and discuss complementary techniques for understanding generalizatio n in such settings.

Evaluating open-ended responses from large audio language models (LALMs) is challenging because human annotators often genuinely disagree on answer correctness due to multiple valid interpretations, partial correctness, and subjective judgment. Traditional metrics reporting only mean scores fail to capture this uncertainty. We present ORCA (Open-ended Response Correctness Assessment), a framework that models the variability in human judgments using Beta distributions to predict both expected correctness and uncertainty. Our three-stage annotation framework combines human judgment with structured feedback and iterative refinement to simultaneously curate training data and improve benchmark quality. We collected 11,721 annotations across 3,580 question-answer pairs from 15 LALMs on two audio QA benchmarks, achieving inter-annotator agreement of 0.82 (Krippendorff's alpha). ORCA achieves 0.91 Spearman correlation with mean human judgments, matching or outperforming LLM-judge baselines while providing uncertainty estimates and requiring significantly less compute. We release our models, code, and curated dataset.

Many indoor workspaces are quasi-static: global layout is stable but local semantics change continually, producing repetitive geometry, dynamic clutter, and perceptual noise that defeat vision-based localization. We present ShelfAware, a semantic particle filter for robust global localization that treats scene semantics as statistical evidence over object categories rather than fixed landmarks. ShelfAware fuses a depth likelihood with a category-centric semantic similarity and uses a precomputed bank of semantic viewpoints to perform inverse semantic proposals inside MCL, yielding fast, targeted hypothesis generation on low-cost, vision-only hardware. Across 100 global-localization trials spanning four conditions (cart-mounted, wearable, dynamic obstacles, and sparse semantics) in a semantically dense, retail environment, ShelfAware achieves a 96% success rate (vs. 22% MCL and 10% AMCL) with a mean time-to-convergence of 1.91s, attains the lowest translational RMSE in all conditions, and maintains stable tracking in 80% of tested sequences, all while running in real time on a consumer laptop-class platform. By modeling semantics distributionally at the category level and leveraging inverse proposals, ShelfAware resolves geometric aliasing and semantic drift common to quasi-static domains. Because the method requires only vision sensors and VIO, it integrates as an infrastructure-free building block for mobile robots in warehouses, labs, and retail settings; as a representative application, it also supports the creation of assistive devices providing start-anytime, shared-control assistive navigation for people with visual impairments.

Accurate precipitation forecasting is essential for hydrometeorological risk management, especially for anticipating extreme rainfall that can lead to flash flooding and infrastructure damage. This study introduces a diffusion-based deep learning (DL) framework that systematically compares three residual prediction strategies differing only in their input sources: (1) a fully data-driven model using only past observations from the Multi-Radar Multi-Sensor (MRMS) system, (2) a corrective model using only forecasts from the High-Resolution Rapid Refresh (HRRR) numerical weather prediction system, and (3) a hybrid model integrating both MRMS and selected HRRR forecast variables. By evaluating these approaches under a unified setup, we provide a clearer understanding of how each data source contributes to predictive skill over the Continental United States (CONUS). Forecasts are produced at 1-km spatial resolution, beginning with direct 1-hour predictions and extending to 12 hours using autoregressive rollouts. Performance is evaluated using both CONUS-wide and region-specific metrics that assess overall performance and skill at extreme rainfall thresholds. Across all lead times, our DL framework consistently outperforms the HRRR baseline in pixel-wise and spatiostatistical metrics. The hybrid model performs best at the shortest lead time, while the HRRR-corrective model outperforms others at longer lead times, maintaining high skill through 12 hours. To assess reliability, we incorporate calibrated uncertainty quantification tailored to the residual learning setup. These gains, particularly at longer lead times, are critical for emergency preparedness, where modest increases in forecast horizon can improve decision-making. This work advances DL-based precipitation forecasting by enhancing predictive skill, reliability, and applicability across regions.

The significant increase in software production, driven by the acceleration of development cycles over the past two decades, has led to a steady rise in software vulnerabilities, as shown by statistics published yearly by the CVE program. The automation of the source code vulnerability detection (CVD) process has thus become essential, and several methods have been proposed ranging from the well established program analysis techniques to the more recent AI-based methods. Our research investigates Large Language Models (LLMs), which are considered among the most performant AI models to date, for the CVD task. The objective is to study their performance and apply different state-of-the-art techniques to enhance their effectiveness for this task. We explore various fine-tuning and prompt engineering settings. We particularly suggest one novel approach for fine-tuning LLMs which we call Double Fine-tuning, and also test the understudied Test-Time fine-tuning approach. We leverage the recent open-source Llama-3.1 8B, with source code samples extracted from BigVul and PrimeVul datasets. Our conclusions highlight the importance of fine-tuning to resolve the task, the performance of Double tuning, as well as the potential of Llama models for CVD. Though prompting proved ineffective, Retrieval augmented generation (RAG) performed relatively well as an example selection technique. Overall, some of our research questions have been answered, and many are still on hold, which leaves us many future work perspectives. Code repository is available here: https://github.com/DynaSoumhaneOuchebara/Llama-based-vulnerability-detection.