Enkelejda Kasneci

Michael Kirchhof, Enkelejda Kasneci, Seong Joon Oh · apple-ml

Contrastively trained encoders have recently been proven to invert the data-generating process: they encode each input, e.g., an image, into the true latent vector that generated the image (Zimmermann et al., 2021). However, real-world observations often have inherent ambiguities. For instance, images may be blurred or only show a 2D view of a 3D object, so multiple latents could have generated them. This makes the true posterior for the latent vector probabilistic with heteroscedastic uncertainty. In this setup, we extend the common InfoNCE objective and encoders to predict latent distributions instead of points. We prove that these distributions recover the correct posteriors of the data-generating process, including its level of aleatoric uncertainty, up to a rotation of the latent space. In addition to providing calibrated uncertainty estimates, these posteriors allow the computation of credible intervals in image retrieval. They comprise images with the same latent as a given query, subject to its uncertainty. Code is available at https://github.com/mkirchhof/Probabilistic_Contrastive_Learning

Michael Kirchhof, Karsten Roth, Zeynep Akata et al. · apple-ml

Proxy-based Deep Metric Learning (DML) learns deep representations by embedding images close to their class representatives (proxies), commonly with respect to the angle between them. However, this disregards the embedding norm, which can carry additional beneficial context such as class- or image-intrinsic uncertainty. In addition, proxy-based DML struggles to learn class-internal structures. To address both issues at once, we introduce non-isotropic probabilistic proxy-based DML. We model images as directional von Mises-Fisher (vMF) distributions on the hypersphere that can reflect image-intrinsic uncertainties. Further, we derive non-isotropic von Mises-Fisher (nivMF) distributions for class proxies to better represent complex class-specific variances. To measure the proxy-to-image distance between these models, we develop and investigate multiple distribution-to-point and distribution-to-distribution metrics. Each framework choice is motivated by a set of ablational studies, which showcase beneficial properties of our probabilistic approach to proxy-based DML, such as uncertainty-awareness, better-behaved gradients during training, and overall improved generalization performance. The latter is especially reflected in the competitive performance on the standard DML benchmarks, where our approach compares favorably, suggesting that existing proxy-based DML can significantly benefit from a more probabilistic treatment. Code is available at github.com/ExplainableML/Probabilistic_Deep_Metric_Learning.

Michael Kirchhof, Bálint Mucsányi, Seong Joon Oh et al. · apple-ml

Representation learning has significantly driven the field to develop pretrained models that can act as a valuable starting point when transferring to new datasets. With the rising demand for reliable machine learning and uncertainty quantification, there is a need for pretrained models that not only provide embeddings but also transferable uncertainty estimates. To guide the development of such models, we propose the Uncertainty-aware Representation Learning (URL) benchmark. Besides the transferability of the representations, it also measures the zero-shot transferability of the uncertainty estimate using a novel metric. We apply URL to evaluate eleven uncertainty quantifiers that are pretrained on ImageNet and transferred to eight downstream datasets. We find that approaches that focus on the uncertainty of the representation itself or estimate the prediction risk directly outperform those that are based on the probabilities of upstream classes. Yet, achieving transferable uncertainty quantification remains an open challenge. Our findings indicate that it is not necessarily in conflict with traditional representation learning goals. Code is provided under https://github.com/mkirchhof/url .

Tobias Leemann, Michael Kirchhof, Yao Rong et al. · apple-ml

Interest in understanding and factorizing learned embedding spaces through conceptual explanations is steadily growing. When no human concept labels are available, concept discovery methods search trained embedding spaces for interpretable concepts like object shape or color that can provide post-hoc explanations for decisions. Unlike previous work, we argue that concept discovery should be identifiable, meaning that a number of known concepts can be provably recovered to guarantee reliability of the explanations. As a starting point, we explicitly make the connection between concept discovery and classical methods like Principal Component Analysis and Independent Component Analysis by showing that they can recover independent concepts under non-Gaussian distributions. For dependent concepts, we propose two novel approaches that exploit functional compositionality properties of image-generating processes. Our provably identifiable concept discovery methods substantially outperform competitors on a battery of experiments including hundreds of trained models and dependent concepts, where they exhibit up to 29 % better alignment with the ground truth. Our results highlight the strict conditions under which reliable concept discovery without human labels can be guaranteed and provide a formal foundation for the domain. Our code is available online.

Kathrin Seßler, Yao Rong, Emek Gözlüklü et al.

The use of Large Language Models (LLMs) in mathematical reasoning has become a cornerstone of related research, demonstrating the intelligence of these models and enabling potential practical applications through their advanced performance, such as in educational settings. Despite the variety of datasets and in-context learning algorithms designed to improve the ability of LLMs to automate mathematical problem solving, the lack of comprehensive benchmarking across different datasets makes it complicated to select an appropriate model for specific tasks. In this project, we present a benchmark that fairly compares seven state-of-the-art in-context learning algorithms for mathematical problem solving across five widely used mathematical datasets on four powerful foundation models. Furthermore, we explore the trade-off between efficiency and performance, highlighting the practical applications of LLMs for mathematical reasoning. Our results indicate that larger foundation models like GPT-4o and LLaMA 3-70B can solve mathematical reasoning independently from the concrete prompting strategy, while for smaller models the in-context learning approach significantly influences the performance. Moreover, the optimal prompt depends on the chosen foundation model. We open-source our benchmark code to support the integration of additional models in future research.

Franziska Kaltenberger, Wei-Ling Chen, Enkeleda Thaqi et al.

Remote and webcam-based eye tracking in multi-line reading suffers from various noise factors and layout ambiguity, precisely where real-time reading support needs reliable, per-fixation line assignment. Prior work largely addresses this challenge post hoc or by restricting behavior (e.g., disallowing re-reading), undermining interactive use. We propose CONF-LA (Confidence-score-based Online Fixation-to-Line Assignment), a principled, low-latency approach that integrates knowledge about reading behavior and Gaussian line likelihoods over fixations to compute a posterior-line-score and defers assignments when uncertainty is high. Evaluated on existing open-source data, CONF-LA demonstrates stable performance in post hoc analysis and closes the online-offline gap (1-2 %) with a mean per-fixation latency of 0.348 ms. Our approach exhibits particular invariance toward regressions, yielding significant improvement in ad hoc median accuracies on children data (approx. 95 %) over all tested algorithms. We encourage further research in this direction and discuss possibilities for future development.

Yao Rong, Guanchu Wang, Qizhang Feng et al.

As Graph Neural Networks (GNNs) have been widely used in real-world applications, model explanations are required not only by users but also by legal regulations. However, simultaneously achieving high fidelity and low computational costs in generating explanations has been a challenge for current methods. In this work, we propose a framework of GNN explanation named LeArn Removal-based Attribution (LARA) to address this problem. Specifically, we introduce removal-based attribution and demonstrate its substantiated link to interpretability fidelity theoretically and experimentally. The explainer in LARA learns to generate removal-based attribution which enables providing explanations with high fidelity. A strategy of subgraph sampling is designed in LARA to improve the scalability of the training process. In the deployment, LARA can efficiently generate the explanation through a feed-forward pass. We benchmark our approach with other state-of-the-art GNN explanation methods on six datasets. Results highlight the effectiveness of our framework regarding both efficiency and fidelity. In particular, LARA is 3.5 times faster and achieves higher fidelity than the state-of-the-art method on the large dataset ogbn-arxiv (more than 160K nodes and 1M edges), showing its great potential in real-world applications. Our source code is available at https://anonymous.4open.science/r/LARA-10D8/README.md.

Chen Xin, Andreas Hartel, Enkelejda Kasneci

Accurate real-time object detection is vital across numerous industrial applications, from safety monitoring to quality control. Traditional approaches, however, are hindered by arduous manual annotation and data collection, struggling to adapt to ever-changing environments and novel target objects. To address these limitations, this paper presents DART, an innovative automated end-to-end pipeline that revolutionizes object detection workflows from data collection to model evaluation. It eliminates the need for laborious human labeling and extensive data collection while achieving outstanding accuracy across diverse scenarios. DART encompasses four key stages: (1) Data Diversification using subject-driven image generation (DreamBooth with SDXL), (2) Annotation via open-vocabulary object detection (Grounding DINO) to generate bounding box and class labels, (3) Review of generated images and pseudo-labels by large multimodal models (InternVL-1.5 and GPT-4o) to guarantee credibility, and (4) Training of real-time object detectors (YOLOv8 and YOLOv10) using the verified data. We apply DART to a self-collected dataset of construction machines named Liebherr Product, which contains over 15K high-quality images across 23 categories. The current instantiation of DART significantly increases average precision (AP) from 0.064 to 0.832. Its modular design ensures easy exchangeability and extensibility, allowing for future algorithm upgrades, seamless integration of new object categories, and adaptability to customized environments without manual labeling and additional data collection. The code and dataset are released at https://github.com/chen-xin-94/DART.

Yao Rong, Tobias Leemann, Thai-trang Nguyen et al.

Explainable AI (XAI) is widely viewed as a sine qua non for ever-expanding AI research. A better understanding of the needs of XAI users, as well as human-centered evaluations of explainable models are both a necessity and a challenge. In this paper, we explore how HCI and AI researchers conduct user studies in XAI applications based on a systematic literature review. After identifying and thoroughly analyzing 97core papers with human-based XAI evaluations over the past five years, we categorize them along the measured characteristics of explanatory methods, namely trust, understanding, usability, and human-AI collaboration performance. Our research shows that XAI is spreading more rapidly in certain application domains, such as recommender systems than in others, but that user evaluations are still rather sparse and incorporate hardly any insights from cognitive or social sciences. Based on a comprehensive discussion of best practices, i.e., common models, design choices, and measures in user studies, we propose practical guidelines on designing and conducting user studies for XAI researchers and practitioners. Lastly, this survey also highlights several open research directions, particularly linking psychological science and human-centered XAI.

Hasan Tarik Akbaba, Efe Bozkir, Anna Puhl et al.

Extended Reality (XR) offers transformative potential for industrial support, training, and maintenance; yet, widespread adoption lags despite demonstrated occupational value and hardware maturity. Organizations successfully implement XR in isolated pilots, yet struggle to scale these into sustained operational deployment, a phenomenon we characterize as the ``Pilot Trap.'' This study examines this phenomenon through a qualitative ecosystem analysis of 17 expert interviews across technology providers, solution integrators, and industrial adopters. We identify a ``Great Inversion'' in adoption barriers: critical constraints have shifted from technological maturity to organizational readiness (e.g., change management, key performance indicator alignment, and political resistance). While hardware ergonomics and usability remain relevant, our findings indicate that systemic misalignments between stakeholder incentives are the primary cause of friction preventing enterprise integration. We conclude that successful industrial XR adoption requires a shift from technology-centric piloting to a problem-first, organizational transformation approach, necessitating explicit ecosystem-level coordination.

Arne Bewersdorff, Kathrin Seßler, Armin Baur et al.

Identifying logical errors in complex, incomplete or even contradictory and overall heterogeneous data like students' experimentation protocols is challenging. Recognizing the limitations of current evaluation methods, we investigate the potential of Large Language Models (LLMs) for automatically identifying student errors and streamlining teacher assessments. Our aim is to provide a foundation for productive, personalized feedback. Using a dataset of 65 student protocols, an Artificial Intelligence (AI) system based on the GPT-3.5 and GPT-4 series was developed and tested against human raters. Our results indicate varying levels of accuracy in error detection between the AI system and human raters. The AI system can accurately identify many fundamental student errors, for instance, the AI system identifies when a student is focusing the hypothesis not on the dependent variable but solely on an expected observation (acc. = 0.90), when a student modifies the trials in an ongoing investigation (acc. = 1), and whether a student is conducting valid test trials (acc. = 0.82) reliably. The identification of other, usually more complex errors, like whether a student conducts a valid control trial (acc. = .60), poses a greater challenge. This research explores not only the utility of AI in educational settings, but also contributes to the understanding of the capabilities of LLMs in error detection in inquiry-based learning like experimentation.

Virmarie Maquiling, Sean Anthony Byrne, Diederick C. Niehorster et al.

The advent of foundation models signals a new era in artificial intelligence. The Segment Anything Model (SAM) is the first foundation model for image segmentation. In this study, we evaluate SAM's ability to segment features from eye images recorded in virtual reality setups. The increasing requirement for annotated eye-image datasets presents a significant opportunity for SAM to redefine the landscape of data annotation in gaze estimation. Our investigation centers on SAM's zero-shot learning abilities and the effectiveness of prompts like bounding boxes or point clicks. Our results are consistent with studies in other domains, demonstrating that SAM's segmentation effectiveness can be on-par with specialized models depending on the feature, with prompts improving its performance, evidenced by an IoU of 93.34% for pupil segmentation in one dataset. Foundation models like SAM could revolutionize gaze estimation by enabling quick and easy image segmentation, reducing reliance on specialized models and extensive manual annotation.

Sean Anthony Byrne, Virmarie Maquiling, Marcus Nyström et al.

Deep learning has bolstered gaze estimation techniques, but real-world deployment has been impeded by inadequate training datasets. This problem is exacerbated by both hardware-induced variations in eye images and inherent biological differences across the recorded participants, leading to both feature and pixel-level variance that hinders the generalizability of models trained on specific datasets. While synthetic datasets can be a solution, their creation is both time and resource-intensive. To address this problem, we present a framework called Light Eyes or "LEyes" which, unlike conventional photorealistic methods, only models key image features required for video-based eye tracking using simple light distributions. LEyes facilitates easy configuration for training neural networks across diverse gaze-estimation tasks. We demonstrate that models trained using LEyes are consistently on-par or outperform other state-of-the-art algorithms in terms of pupil and CR localization across well-known datasets. In addition, a LEyes trained model outperforms the industry standard eye tracker using significantly more cost-effective hardware. Going forward, we are confident that LEyes will revolutionize synthetic data generation for gaze estimation models, and lead to significant improvements of the next generation video-based eye trackers.

Ka Hei Carrie Lau, Efe Bozkir, Hong Gao et al.

This paper explores the innovative application of Large Language Models (LLMs) in Virtual Reality (VR) environments to promote heritage education, focusing on traditional Scottish curling presented in the game ``Scottish Bonspiel VR''. Our study compares the effectiveness of LLM-based chatbots with pre-defined scripted chatbots, evaluating key criteria such as usability, user engagement, and learning outcomes. The results show that LLM-based chatbots significantly improve interactivity and engagement, creating a more dynamic and immersive learning environment. This integration helps document and preserve cultural heritage and enhances dissemination processes, which are crucial for safeguarding intangible cultural heritage (ICH) amid environmental changes. Furthermore, the study highlights the potential of novel technologies in education to provide immersive experiences that foster a deeper appreciation of cultural heritage. These findings support the wider application of LLMs and VR in cultural education to address global challenges and promote sustainable practices to preserve and enhance cultural heritage.

Daniel Weber, Wolfgang Fuhl, Andreas Zell et al.

In human-robot collaboration, one challenging task is to teach a robot new yet unknown objects enabling it to interact with them. Thereby, gaze can contain valuable information. We investigate if it is possible to detect objects (object or no object) merely from gaze data and determine their bounding box parameters. For this purpose, we explore different sizes of temporal windows, which serve as a basis for the computation of heatmaps, i.e., the spatial distribution of the gaze data. Additionally, we analyze different grid sizes of these heatmaps, and demonstrate the functionality in a proof of concept using different machine learning techniques. Our method is characterized by its speed and resource efficiency compared to conventional object detectors. In order to generate the required data, we conducted a study with five subjects who could move freely and thus, turn towards arbitrary objects. This way, we chose a scenario for our data collection that is as realistic as possible. Since the subjects move while facing objects, the heatmaps also contain gaze data trajectories, complicating the detection and parameter regression. We make our data set publicly available to the research community for download.

Sean Anthony Byrne, Marcus Nyström, Virmarie Maquiling et al.

We present a deep learning method for accurately localizing the center of a single corneal reflection (CR) in an eye image. Unlike previous approaches, we use a convolutional neural network (CNN) that was trained solely using simulated data. Using only simulated data has the benefit of completely sidestepping the time-consuming process of manual annotation that is required for supervised training on real eye images. To systematically evaluate the accuracy of our method, we first tested it on images with simulated CRs placed on different backgrounds and embedded in varying levels of noise. Second, we tested the method on high-quality videos captured from real eyes. Our method outperformed state-of-the-art algorithmic methods on real eye images with a 35% reduction in terms of spatial precision, and performed on par with state-of-the-art on simulated images in terms of spatial accuracy.We conclude that our method provides a precise method for CR center localization and provides a solution to the data availability problem which is one of the important common roadblocks in the development of deep learning models for gaze estimation. Due to the superior CR center localization and ease of application, our method has the potential to improve the accuracy and precision of CR-based eye trackers

Yao Rong, Naemi-Rebecca Kassautzki, Wolfgang Fuhl et al.

Human drivers use their attentional mechanisms to focus on critical objects and make decisions while driving. As human attention can be revealed from gaze data, capturing and analyzing gaze information has emerged in recent years to benefit autonomous driving technology. Previous works in this context have primarily aimed at predicting "where" human drivers look at and lack knowledge of "what" objects drivers focus on. Our work bridges the gap between pixel-level and object-level attention prediction. Specifically, we propose to integrate an attention prediction module into a pretrained object detection framework and predict the attention in a grid-based style. Furthermore, critical objects are recognized based on predicted attended-to areas. We evaluate our proposed method on two driver attention datasets, BDD-A and DR(eye)VE. Our framework achieves competitive state-of-the-art performance in the attention prediction on both pixel-level and object-level but is far more efficient (75.3 GFLOPs less) in computation.

Anna Bodonhelyi, Enkeleda Thaqi, Süleyman Özdel et al.

In online education, innovative tools are crucial for enhancing learning outcomes. SAM (Study with AI Mentor) is an advanced platform that integrates educational videos with a context-aware chat interface powered by large language models. SAM encourages students to ask questions and explore unclear concepts in real time, offering personalized, context-specific assistance, including explanations of formulas, slides, and images. We evaluated SAM in two studies: one with 25 university students and another with 80 crowdsourced participants, using pre- and post-knowledge tests to compare a group using SAM and a control group. The results demonstrated that SAM users achieved greater knowledge gains specifically for younger learners and individuals in flexible working environments, such as students, supported by a 97.6% accuracy rate in the chatbot's responses. Participants also provided positive feedback on SAM's usability and effectiveness. SAM's proactive approach to learning not only enhances learning outcomes but also empowers students to take full ownership of their educational experience, representing a promising future direction for online learning tools.

Philipp Hallgarten, David Bethge, Ozan Özdenizci et al.

Limited availability of labeled physiological data often prohibits the use of powerful supervised deep learning models in the biomedical machine intelligence domain. We approach this problem and propose a novel encoding framework that relies on self-supervised learning with momentum contrast to learn representations from multivariate time-series of various physiological domains without needing labels. Our model uses a transformer architecture that can be easily adapted to classification problems by optimizing a linear output classification layer. We experimentally evaluate our framework using two publicly available physiological datasets from different domains, i.e., human activity recognition from embedded inertial sensory and emotion recognition from electroencephalography. We show that our self-supervised learning approach can indeed learn discriminative features which can be exploited in downstream classification tasks. Our work enables the development of domain-agnostic intelligent systems that can effectively analyze multivariate time-series data from physiological domains.

Dongyang Guo, Yasmeen Abdrabou, Enkelejda Kasneci

Gaze event detection is fundamental to vision science, human-computer interaction, and applied analytics. However, current workflows often require specialized programming knowledge and careful handling of heterogeneous raw data formats. Classical detectors such as I-VT and I-DT are effective but highly sensitive to preprocessing and parameterization, limiting their usability outside specialized laboratories. This work introduces a code-free, large language model (LLM)-driven pipeline that converts natural language instructions into an end-to-end analysis. The system (1) inspects raw eye-tracking files to infer structure and metadata, (2) generates executable routines for data cleaning and detector implementation from concise user prompts, (3) applies the generated detector to label fixations and saccades, and (4) returns results and explanatory reports, and allows users to iteratively optimize their code by editing the prompt. Evaluated on public benchmarks, the approach achieves accuracy comparable to traditional methods while substantially reducing technical overhead. The framework lowers barriers to entry for eye-tracking research, providing a flexible and accessible alternative to code-intensive workflows.

Virmarie Maquiling, Yasmeen Abdrabou, Enkelejda Kasneci

Corneal reflection (glint) detection plays an important role in pupil-corneal reflection (P-CR) eye tracking, but in practice it is often handled as heuristics embedded within larger systems, making reproducibility difficult across hardware setups. We introduce a 2D geometry-driven, constellation-based pipeline for mulit-glint detection and matching, focusing on reproducibility and clear evaluation. Inspired by lost-in-space star identification, we treat glints as structured constellations rather than independent blobs. We propose a Similarity-Layout Alignment (SLA) procedure which adapts constellation matching to the specific constraints of multi-LED eye tracking. The framework brings together controlled over-detection, adaptive candidate fallback, appearance-aware scoring, and optional semantic layout priors while keeping detection and correspondence explicitly separated. Evaluated on a public multi-LED dataset, the system provides stable identity-preserving correspondence under noisy conditions. We release code, presets, and evaluation scripts to enable transparent replication, comparison, and dataset annotation.

Virmarie Maquiling, Yasmeen Abdrabou, Enkelejda Kasneci

Vergence is widely used as a proxy for depth perception and spatial attention in immersive and real-world eye-tracking studies. In this paper, we investigate how pupil size artefacts affect vergence estimates during real physical depth viewing with a head-mounted eye tracker. Using a beamsplitter setup with physically near and far targets, we elicited controlled convergent and divergent eye movements under static, luminance-modulated, and blockwise fixation conditions. Near and far targets were reliably separable in vergence angle across participants. However, pupil-vergence coupling varied substantially across individuals and conditions. Static illumination produced large inter-participant variability, while luminance modulation reduced this spread, yielding more clustered estimates. Blockwise and audio-cued recordings further showed that pupil-vergence coupling persists even without visual depth onsets. These results suggest that pupil size fluctuations can systematically influence vergence estimates, and that controlled viewing conditions can reduce--but not eliminate--this effect.

Anna Bodonhelyi, Augustin Curinier, Babette Bühler et al.

Detecting mind wandering is crucial in online education, and it occurs 30% of the time, as it directly impacts learners' retention, comprehension, and overall success in self-directed learning environments. Integrating automated detection algorithms enables the deployment of targeted interventions within adaptive learning environments, paving the way for more responsive and personalized educational systems. However, progress is hampered by a lack of coherent frameworks for identifying mind wandering in online environments. This work presents a comprehensive systematic review and benchmark of mind wandering detection across 14 datasets covering EEG, facial video, eye tracking, and physiological signals in educational settings, motivated by the challenges in achieving reliable detection and the inconsistency of results across studies caused by variations in models, preprocessing approaches, and evaluation metrics. We implemented a generalizable preprocessing and feature extraction pipeline tailored to each modality, ensuring fair comparison across diverse experimental paradigms. 13 traditional machine learning and neural network models, including federated learning approaches, were evaluated on each dataset. In a novel ablation study, we explored mind wandering detection from post-probe data, motivated by findings that learners often re-engage with material after mind wandering episodes through re-reading or re-watching. Results highlight the potential and limitations of different modalities and classifiers for mind wandering detection, and point to new opportunities for supporting online learning. All code and preprocessing scripts are made openly available to support reproducibility and future research.

Ka Hei Carrie Lau, Philipp Stark, Efe Bozkir et al.

Artificial intelligence is increasingly used in hiring, raising concerns about how applicants perceive these systems. While prior work on algorithmic fairness has emphasized technical bias mitigation, little is known about how avatar identity cues influence applicants' justice attributions in an interview context. We conducted a crowdsourcing study with 215 participants who completed an interview with photorealistic AI avatars varied in phenotypic traits (race and sex), followed by a standardized rejection. Using self-reports, sentiment analysis, and eye tracking, we measured perceptions of trust, fairness, and bias. Results show that racial mismatch heightened perceptions of ethnic bias, while partial match (sharing only one identity) reduced fairness judgments compared to both full and no match. This work extends the Computers-Are-Social-Actors paradigm by demonstrating that avatar appearances shape justicerelated evaluations of AI. We contribute to HCI by revealing how identity cues influence fairness attributions and offer actionable insights for designing equitable AI interview systems.

Anna Bodonhelyi, Mengdi Wang, Efe Bozkir et al.

Since the COVID-19 pandemic, online courses have expanded access to education, yet the absence of direct instructor support challenges learners' ability to self-regulate attention and engagement. Mind wandering and disengagement can be detrimental to learning outcomes, making their automated detection via video-based indicators a promising approach for real-time learner support. However, machine learning-based approaches often require sharing sensitive data, raising privacy concerns. Federated learning offers a privacy-preserving alternative by enabling decentralized model training while also distributing computational load. We propose a framework exploiting cross-device federated learning to address different manifestations of behavioral and cognitive disengagement during remote learning, specifically behavioral disengagement, mind wandering, and boredom. We fit video-based cognitive disengagement detection models using facial expressions and gaze features. By adopting federated learning, we safeguard users' data privacy through privacy-by-design and introduce a novel solution with the potential for real-time learner support. We further address challenges posed by eyeglasses by incorporating related features, enhancing overall model performance. To validate the performance of our approach, we conduct extensive experiments on five datasets and benchmark multiple federated learning algorithms. Our results show great promise for privacy-preserving educational technologies promoting learner engagement.

Efe Bozkir, Enkelejda Kasneci

The rapid development of generative artificial intelligence (AI) and large language models (LLMs), and the availability of services that make them accessible, have led the general public to begin incorporating them into everyday life. The extended reality (XR) community has also sought to integrate LLMs, particularly in the form of conversational agents, to enhance user experience and task efficiency. When interacting with such conversational agents, users may easily disclose sensitive information due to the naturalistic flow of the conversations, and combining such conversational data with fine-grained sensor data may lead to novel privacy issues. To address these issues, a user-centric understanding of technology acceptance and concerns is essential. Therefore, to this end, we conducted a large-scale crowdsourcing study with 1036 participants, examining user decision-making processes regarding LLM-powered conversational agents in XR, across factors of XR setting type, speech interaction type, and data processing location. We found that while users generally accept these technologies, they express concerns related to security, privacy, social implications, and trust. Our results suggest that familiarity plays a crucial role, as daily generative AI use is associated with greater acceptance. In contrast, previous ownership of XR devices is linked to less acceptance, possibly due to existing familiarity with the settings. We also found that men report higher acceptance with fewer concerns than women. Regarding data type sensitivity, location data elicited the most significant concern, while body temperature and virtual object states were considered least sensitive. Overall, our study highlights the importance of practitioners effectively communicating their measures to users, who may remain distrustful. We conclude with implications and recommendations for LLM-powered XR.

Aditi Bhalla, Christian Hellert, Enkelejda Kasneci

Driver distraction remains a leading cause of road traffic accidents, contributing to thousands of fatalities annually across the globe. While deep learning-based driver activity recognition methods have shown promise in detecting such distractions, their effectiveness in real-world deployments is hindered by two critical challenges: variations in camera viewpoints (cross-view) and domain shifts such as change in sensor modality or environment. Existing methods typically address either cross-view generalization or unsupervised domain adaptation in isolation, leaving a gap in the robust and scalable deployment of models across diverse vehicle configurations. In this work, we propose a novel two-phase cross-view, cross-modal unsupervised domain adaptation framework that addresses these challenges jointly on real-time driver monitoring data. In the first phase, we learn view-invariant and action-discriminative features within a single modality using contrastive learning on multi-view data. In the second phase, we perform domain adaptation to a new modality using information bottleneck loss without requiring any labeled data from the new domain. We evaluate our approach using state-of-the art video transformers (Video Swin, MViT) and multi modal driver activity dataset called Drive&Act, demonstrating that our joint framework improves top-1 accuracy on RGB video data by almost 50% compared to a supervised contrastive learning-based cross-view method, and outperforms unsupervised domain adaptation-only methods by up to 5%, using the same video transformer backbone.

Kathrin Seßler, Maurice Fürstenberg, Babette Bühler et al.

The manual assessment and grading of student writing is a time-consuming yet critical task for teachers. Recent developments in generative AI, such as large language models, offer potential solutions to facilitate essay-scoring tasks for teachers. In our study, we evaluate the performance and reliability of both open-source and closed-source LLMs in assessing German student essays, comparing their evaluations to those of 37 teachers across 10 pre-defined criteria (i.e., plot logic, expression). A corpus of 20 real-world essays from Year 7 and 8 students was analyzed using five LLMs: GPT-3.5, GPT-4, o1, LLaMA 3-70B, and Mixtral 8x7B, aiming to provide in-depth insights into LLMs' scoring capabilities. Closed-source GPT models outperform open-source models in both internal consistency and alignment with human ratings, particularly excelling in language-related criteria. The novel o1 model outperforms all other LLMs, achieving Spearman's $r = .74$ with human assessments in the overall score, and an internal consistency of $ICC=.80$. These findings indicate that LLM-based assessment can be a useful tool to reduce teacher workload by supporting the evaluation of essays, especially with regard to language-related criteria. However, due to their tendency for higher scores, the models require further refinement to better capture aspects of content quality.

Alexander Zimmer, Yasmeen Abdrabou, Enkelejda Kasneci

Research on video-based eye-tracking has long explored stereo and glint-based methods, yet existing wearable eye trackers - both commercial and open-source - offer limited flexibility for algorithm development and comparative evaluation. We present an affordable, wearable stereo eye-tracking platform built from off-the-shelf and 3D-printable components that explicitly targets this gap. The system combines four infrared eye cameras, infrared illumination, an optional scene camera, and software support for calibration and synchronized data acquisition. By design, the platform supports multiple eye-tracking paradigms, including stereo, glint-based, and binocular approaches, within a single hardware configuration. Rather than optimizing for end-user robustness, the platform prioritizes modularity and extensibility for research use. This paper focuses on the hardware architecture and calibration pipeline and demonstrates the feasibility of the approach using a prototype implementation. All hardware designs and documentation are made openly available.

Michael Kirchhof, Mark Collier, Seong Joon Oh et al. · apple-ml

Accurate uncertainty estimation is vital to trustworthy machine learning, yet uncertainties typically have to be learned for each task anew. This work introduces the first pretrained uncertainty modules for vision models. Similar to standard pretraining this enables the zero-shot transfer of uncertainties learned on a large pretraining dataset to specialized downstream datasets. We enable our large-scale pretraining on ImageNet-21k by solving a gradient conflict in previous uncertainty modules and accelerating the training by up to 180x. We find that the pretrained uncertainties generalize to unseen datasets. In scrutinizing the learned uncertainties, we find that they capture aleatoric uncertainty, disentangled from epistemic components. We demonstrate that this enables safe retrieval and uncertainty-aware dataset visualization. To encourage applications to further problems and domains, we release all pretrained checkpoints and code under https://github.com/mkirchhof/url .

Kadir Burak Buldu, Süleyman Özdel, Ka Hei Carrie Lau et al.

Recent developments in computer graphics, machine learning, and sensor technologies enable numerous opportunities for extended reality (XR) setups for everyday life, from skills training to entertainment. With large corporations offering affordable consumer-grade head-mounted displays (HMDs), XR will likely become pervasive, and HMDs will develop as personal devices like smartphones and tablets. However, having intelligent spaces and naturalistic interactions in XR is as important as technological advances so that users grow their engagement in virtual and augmented spaces. To this end, large language model (LLM)--powered non-player characters (NPCs) with speech-to-text (STT) and text-to-speech (TTS) models bring significant advantages over conventional or pre-scripted NPCs for facilitating more natural conversational user interfaces (CUIs) in XR. This paper provides the community with an open-source, customizable, extendable, and privacy-aware Unity package, CUIfy, that facilitates speech-based NPC-user interaction with widely used LLMs, STT, and TTS models. Our package also supports multiple LLM-powered NPCs per environment and minimizes latency between different computational models through streaming to achieve usable interactions between users and NPCs. We publish our source code in the following repository: https://gitlab.lrz.de/hctl/cuify

Zilong Zhao, Yao Rong, Dongyang Guo et al.

Using Large Language Models for complex mathematical reasoning is difficult, primarily due to the complexity of multi-step reasoning. The main challenges of this process include (1) selecting critical intermediate results to advance the procedure, and (2) limited exploration of potential solutions. To address these issues, we introduce a novel algorithm, namely Stepwise Self-Consistent Chain-of-Thought (SSC-CoT). SSC-CoT employs a strategy of selecting intermediate steps based on the intersection of various reasoning chains. Additionally, SSC-CoT enables the model to discover critical intermediate steps by querying a knowledge graph comprising relevant domain knowledge. To validate SSC-CoT, we present a new dataset, TriMaster100, tailored for complex trigonometry problems. This dataset contains 100 questions, with each solution broken down into scored intermediate steps, facilitating a comprehensive evaluation of the mathematical reasoning process. On TriMaster100, SSC-CoT triples the effectiveness of the state-of-the-art methods. Furthermore, we benchmark SSC-CoT on the widely recognized complex mathematical question dataset, MATH level 5, and it surpasses the second-best method by 7.2% in accuracy. Code and the TriMaster100 dataset can be found at: https://github.com/zhao-zilong/ssc-cot.

Enkelejda Kasneci, Gjergji Kasneci

This position paper argues that effective tutoring requires corrective friction: surfacing misconceptions and challenging them supportively to drive conceptual change. Yet preference-aligned LLMs can trade epistemic rigor for agreeableness. We identify a Reasoning-Sycophancy Paradox: models that resist context-switch frame attacks can still capitulate under social-epistemic pressure, especially authority ("my notes say I'm right") and social-affective face-saving ("please don't tell me I'm wrong"). We introduce EduFrameTrap, a tutoring benchmark across math, physics, economics, chemistry, biology, and computer science that varies student confidence and pressure (context-switch, authority, social-affective). Across two frontier LLMs, context-switch failures are comparatively lower for GPT-5.2, while authority and social pressure more often trigger epistemic retreat. In contrast, Claude shows substantial context-switch fragility in this run. Because these failures are hard to judge automatically, we report two-judge disagreement as a reliability signal. We argue benchmarks should measure social-epistemic courage, i.e., supportive but corrective tutoring, and treat kind-but-correct behavior as a safety requirement.

Virmarie Maquiling, Sean Anthony Byrne, Diederick C. Niehorster et al.

We explore the transformative potential of SAM 2, a vision foundation model, in advancing gaze estimation and eye tracking technologies. By significantly reducing annotation time, lowering technical barriers through its ease of deployment, and enhancing segmentation accuracy, SAM 2 addresses critical challenges faced by researchers and practitioners. Utilizing its zero-shot segmentation capabilities with minimal user input-a single click per video-we tested SAM 2 on over 14 million eye images from diverse datasets, including virtual reality setups and the world's largest unified dataset recorded using wearable eye trackers. Remarkably, in pupil segmentation tasks, SAM 2 matches the performance of domain-specific models trained solely on eye images, achieving competitive mean Intersection over Union (mIoU) scores of up to 93% without fine-tuning. Additionally, we provide our code and segmentation masks for these widely used datasets to promote further research.

Chen Xin, Thomas Motz, Andreas Hartel et al.

Real-time object localization on edge devices is fundamental for numerous applications, ranging from surveillance to industrial automation. Traditional frameworks, such as object detection, segmentation, and keypoint detection, struggle in resource-constrained environments, often resulting in substantial target omissions. To address these challenges, we introduce OCDet, a lightweight Object Center Detection framework optimized for edge devices with NPUs. OCDet predicts heatmaps representing object center probabilities and extracts center points through peak identification. Unlike prior methods using fixed Gaussian distribution, we introduce Generalized Centerness (GC) to generate ground truth heatmaps from bounding box annotations, providing finer spatial details without additional manual labeling. Built on NPU-friendly Semantic FPN with MobileNetV4 backbones, OCDet models are trained by our Balanced Continuous Focal Loss (BCFL), which alleviates data imbalance and focuses training on hard negative examples for probability regression tasks. Leveraging the novel Center Alignment Score (CAS) with Hungarian matching, we demonstrate that OCDet consistently outperforms YOLO11 in object center detection, achieving up to 23% higher CAS while requiring 42% fewer parameters, 34% less computation, and 64% lower NPU latency. When compared to keypoint detection frameworks, OCDet achieves substantial CAS improvements up to 186% using identical models. By integrating GC, BCFL, and CAS, OCDet establishes a new paradigm for efficient and robust object center detection on edge devices with NPUs. The code is released at https://github.com/chen-xin-94/ocdet.

Mengdi Wang, Efe Bozkir, Enkelejda Kasneci

Split learning emerges as a promising paradigm for collaborative distributed model training, akin to federated learning, by partitioning neural networks between clients and a server without raw data exchange. However, sequential split learning suffers from poor scalability, while parallel variants like parallel split learning and split federated learning often incur high server resource overhead due to model duplication and aggregation, and generally exhibit reduced model performance and convergence owing to factors like client drift and lag. To address these limitations, we introduce CycleSL, a novel aggregation-free split learning framework that enhances scalability and performance and can be seamlessly integrated with existing methods. Inspired by alternating block coordinate descent, CycleSL treats server-side training as an independent higher-level machine learning task, resampling client-extracted features (smashed data) to mitigate heterogeneity and drift. It then performs cyclical updates, namely optimizing the server model first, followed by client updates using the updated server for gradient computation. We integrate CycleSL into previous algorithms and benchmark them on five publicly available datasets with non-iid data distribution and partial client attendance. Our empirical findings highlight the effectiveness of CycleSL in enhancing model performance. Our source code is available at https://gitlab.lrz.de/hctl/CycleSL.

Yao Rong, Tobias Leemann, Vadim Borisov et al.

With a variety of local feature attribution methods being proposed in recent years, follow-up work suggested several evaluation strategies. To assess the attribution quality across different attribution techniques, the most popular among these evaluation strategies in the image domain use pixel perturbations. However, recent advances discovered that different evaluation strategies produce conflicting rankings of attribution methods and can be prohibitively expensive to compute. In this work, we present an information-theoretic analysis of evaluation strategies based on pixel perturbations. Our findings reveal that the results are strongly affected by information leakage through the shape of the removed pixels as opposed to their actual values. Using our theoretical insights, we propose a novel evaluation framework termed Remove and Debias (ROAD) which offers two contributions: First, it mitigates the impact of the confounders, which entails higher consistency among evaluation strategies. Second, ROAD does not require the computationally expensive retraining step and saves up to 99% in computational costs compared to the state-of-the-art. We release our source code at https://github.com/tleemann/road_evaluation.

Yao Rong, Wenjia Xu, Zeynep Akata et al.

The way humans attend to, process and classify a given image has the potential to vastly benefit the performance of deep learning models. Exploiting where humans are focusing can rectify models when they are deviating from essential features for correct decisions. To validate that human attention contains valuable information for decision-making processes such as fine-grained classification, we compare human attention and model explanations in discovering important features. Towards this goal, we collect human gaze data for the fine-grained classification dataset CUB and build a dataset named CUB-GHA (Gaze-based Human Attention). Furthermore, we propose the Gaze Augmentation Training (GAT) and Knowledge Fusion Network (KFN) to integrate human gaze knowledge into classification models. We implement our proposals in CUB-GHA and the recently released medical dataset CXR-Eye of chest X-ray images, which includes gaze data collected from a radiologist. Our result reveals that integrating human attention knowledge benefits classification effectively, e.g. improving the baseline by 4.38% on CXR. Hence, our work provides not only valuable insights into understanding human attention in fine-grained classification, but also contributes to future research in integrating human gaze with computer vision tasks. CUB-GHA and code are available at https://github.com/yaorong0921/CUB-GHA.

Arne Bewersdorff, Christian Hartmann, Marie Hornberger et al.

The integration of Artificial Intelligence (AI), particularly Large Language Model (LLM)-based systems, in education has shown promise in enhancing teaching and learning experiences. However, the advent of Multimodal Large Language Models (MLLMs) like GPT-4 with vision (GPT-4V), capable of processing multimodal data including text, sound, and visual inputs, opens a new era of enriched, personalized, and interactive learning landscapes in education. Grounded in theory of multimedia learning, this paper explores the transformative role of MLLMs in central aspects of science education by presenting exemplary innovative learning scenarios. Possible applications for MLLMs could range from content creation to tailored support for learning, fostering competencies in scientific practices, and providing assessment and feedback. These scenarios are not limited to text-based and uni-modal formats but can be multimodal, increasing thus personalization, accessibility, and potential learning effectiveness. Besides many opportunities, challenges such as data protection and ethical considerations become more salient, calling for robust frameworks to ensure responsible integration. This paper underscores the necessity for a balanced approach in implementing MLLMs, where the technology complements rather than supplants the educator's role, ensuring thus an effective and ethical use of AI in science education. It calls for further research to explore the nuanced implications of MLLMs on the evolving role of educators and to extend the discourse beyond science education to other disciplines. Through the exploration of potentials, challenges, and future implications, we aim to contribute to a preliminary understanding of the transformative trajectory of MLLMs in science education and beyond.

Efe Bozkir, Süleyman Özdel, Ka Hei Carrie Lau et al.

Advances in artificial intelligence and human-computer interaction will likely lead to extended reality (XR) becoming pervasive. While XR can provide users with interactive, engaging, and immersive experiences, non-player characters are often utilized in pre-scripted and conventional ways. This paper argues for using large language models (LLMs) in XR by embedding them in avatars or as narratives to facilitate inclusion through prompt engineering and fine-tuning the LLMs. We argue that this inclusion will promote diversity for XR use. Furthermore, the versatile conversational capabilities of LLMs will likely increase engagement in XR, helping XR become ubiquitous. Lastly, we speculate that combining the information provided to LLM-powered spaces by users and the biometric data obtained might lead to novel privacy invasions. While exploring potential privacy breaches, examining user privacy concerns and preferences is also essential. Therefore, despite challenges, LLM-powered XR is a promising area with several opportunities.

Ruikun Hou, Tim Fütterer, Babette Bühler et al.

Classroom observation protocols standardize the assessment of teaching effectiveness and facilitate comprehension of classroom interactions. Whereas these protocols offer teachers specific feedback on their teaching practices, the manual coding by human raters is resource-intensive and often unreliable. This has sparked interest in developing AI-driven, cost-effective methods for automating such holistic coding. Our work explores a multimodal approach to automatically estimating encouragement and warmth in classrooms, a key component of the Global Teaching Insights (GTI) study's observation protocol. To this end, we employed facial and speech emotion recognition with sentiment analysis to extract interpretable features from video, audio, and transcript data. The prediction task involved both classification and regression methods. Additionally, in light of recent large language models' remarkable text annotation capabilities, we evaluated ChatGPT's zero-shot performance on this scoring task based on transcripts. We demonstrated our approach on the GTI dataset, comprising 367 16-minute video segments from 92 authentic lesson recordings. The inferences of GPT-4 and the best-trained model yielded correlations of r = .341 and r = .441 with human ratings, respectively. Combining estimates from both models through averaging, an ensemble approach achieved a correlation of r = .513, comparable to human inter-rater reliability. Our model explanation analysis indicated that text sentiment features were the primary contributors to the trained model's decisions. Moreover, GPT-4 could deliver logical and concrete reasoning as potential teacher guidelines. Our findings provide insights into using advanced, multimodal techniques for automated classroom observation, aiming to foster teacher training through frequent and valuable feedback.

Ivo Bueno, Lea Hirlimann, Enkelejda Kasneci

Detecting hate speech in memes is challenging due to their multimodal nature and subtle, culturally grounded cues such as sarcasm and context. While recent vision-language models (VLMs) enable joint reasoning over text and images, end-to-end prompting can be brittle, as a single prediction must resolve target, stance, implicitness, and irony. These challenges are amplified in multilingual settings. We propose a prompted weak supervision (PWS) approach that decomposes meme understanding into targeted, question-based labeling functions with constrained answer options for homophobia and transphobia detection in the LT-EDI 2026 shared task. Using a quantized Qwen3-VLM to extract features by answering targeted questions, our method outperforms direct VLM classification, with substantial gains for Chinese and Hindi, ranking 1st in English, 2nd in Chinese, and 3rd in Hindi. Iterative refinement via error-driven LF expansion and feature pruning reduces redundancy and improves generalization. Our results highlight the effectiveness of prompted weak supervision for multilingual multimodal hate speech detection.

Michael Kirchhof, Gjergji Kasneci, Enkelejda Kasneci · apple-ml

Large-language models (LLMs) and chatbot agents are known to provide wrong outputs at times, and it was recently found that this can never be fully prevented. Hence, uncertainty quantification plays a crucial role, aiming to quantify the level of ambiguity in either one overall number or two numbers for aleatoric and epistemic uncertainty. This position paper argues that this traditional dichotomy of uncertainties is too limited for the open and interactive setup that LLM agents operate in when communicating with a user, and that we need to research avenues that enrich uncertainties in this novel scenario. We review the literature and find that popular definitions of aleatoric and epistemic uncertainties directly contradict each other and lose their meaning in interactive LLM agent settings. Hence, we propose three novel research directions that focus on uncertainties in such human-computer interactions: Underspecification uncertainties, for when users do not provide all information or define the exact task at the first go, interactive learning, to ask follow-up questions and reduce the uncertainty about the current context, and output uncertainties, to utilize the rich language and speech space to express uncertainties as more than mere numbers. We expect that these new ways of dealing with and communicating uncertainties will lead to LLM agent interactions that are more transparent, trustworthy, and intuitive.

Suleyman Ozdel, Yao Rong, Berat Mert Albaba et al. · eth-zurich

Humans utilize their gaze to concentrate on essential information while perceiving and interpreting intentions in videos. Incorporating human gaze into computational algorithms can significantly enhance model performance in video understanding tasks. In this work, we address a challenging and innovative task in video understanding: predicting the actions of an agent in a video based on a partial video. We introduce the Gaze-guided Action Anticipation algorithm, which establishes a visual-semantic graph from the video input. Our method utilizes a Graph Neural Network to recognize the agent's intention and predict the action sequence to fulfill this intention. To assess the efficiency of our approach, we collect a dataset containing household activities generated in the VirtualHome environment, accompanied by human gaze data of viewing videos. Our method outperforms state-of-the-art techniques, achieving a 7\% improvement in accuracy for 18-class intention recognition. This highlights the efficiency of our method in learning important features from human gaze data.

Gjergji Kasneci, Enkelejda Kasneci

Feature engineering is crucial for optimizing machine learning model performance, particularly in tabular data classification tasks. Leveraging advancements in natural language processing, this study presents a systematic approach to enrich tabular datasets with features derived from large language model embeddings. Through a comprehensive ablation study on diverse datasets, we assess the impact of RoBERTa and GPT-2 embeddings on ensemble classifiers, including Random Forest, XGBoost, and CatBoost. Results indicate that integrating embeddings with traditional numerical and categorical features often enhances predictive performance, especially on datasets with class imbalance or limited features and samples, such as UCI Adult, Heart Disease, Titanic, and Pima Indian Diabetes, with improvements particularly notable in XGBoost and CatBoost classifiers. Additionally, feature importance analysis reveals that LLM-derived features frequently rank among the most impactful for the predictions. This study provides a structured approach to embedding-based feature enrichment and illustrates its benefits in ensemble learning for tabular data.

Suleyman Ozdel, Yao Rong, Berat Mert Albaba et al. · eth-zurich

Eye-tracking applications that utilize the human gaze in video understanding tasks have become increasingly important. To effectively automate the process of video analysis based on eye-tracking data, it is important to accurately replicate human gaze behavior. However, this task presents significant challenges due to the inherent complexity and ambiguity of human gaze patterns. In this work, we introduce a novel method for simulating human gaze behavior. Our approach uses a transformer-based reinforcement learning algorithm to train an agent that acts as a human observer, with the primary role of watching videos and simulating human gaze behavior. We employed an eye-tracking dataset gathered from videos generated by the VirtualHome simulator, with a primary focus on activity recognition. Our experimental results demonstrate the effectiveness of our gaze prediction method by highlighting its capability to replicate human gaze behavior and its applicability for downstream tasks where real human-gaze is used as input.

Suleyman Ozdel, Virmarie Maquiling, Kadir Burak Buldu et al.

Reproducibility in eye-tracking research is increasingly important as researchers conduct diverse experiments and seek to validate or replicate findings. However, exact replication remains challenging due to differences in laboratory practices and experimental setups. Inconsistent stimulus presentation can yield divergent metrics from identical oculomotor behavior, yet the stimulus layer remains largely unstandardized. Existing tools often require programming expertise or depend on specific hardware vendors. We introduce VIVA Stimuli, a web-based platform for standardized eye-tracking stimulus presentation. It provides configurable task types, including fixation, smooth pursuit, cognitive load, blink, slippage, content display, and questionnaires within a unified environment. The platform supports any eye-tracking technology, including wearable and screen-based VOG trackers, LFI sensors, and EOG devices. ArUco markers enable synchronization for trackers with scene cameras, while a WebSocket architecture ensures temporal synchronization for those without. A visual experiment flow editor allows protocols to be exported and shared, enabling identical stimulus replication across laboratories.

Suleyman Ozdel, Amr Nader, Yasmeen Abdrabou et al.

With the growing use of eye tracking on VR and mobile platforms, gaze data is increasing. While scanpath comparison is important to gaze behavior analysis, existing methods lack privacy-preserving capabilities for real-world use. We present a garbled-circuit (GC)-based approach enabling secure storage and privacy-preserving scanpath comparison under the semi-honest model. It supports two configurations: (1) a two-party setting where the data owner and processor jointly compute similarity scores without revealing their inputs, and (2) a server-assisted setting where encrypted scanpaths are stored and processed while the data owner remains offline. All decryption and comparison operations are executed inside the GC. Experiments on three eye-tracking datasets evaluate fidelity, runtime, and communication, and show secure results for MultiMatch, ScanMatch, and SubsMatch closely match plaintext outcomes, with manageable runtime and communication overhead. Tests under various network conditions indicate that the design remains feasible for real-world privacy-preserving scanpath analysis and can be extended to other GC-based behavioral algorithms.

Duru Paker, Suleyman Ozdel, Enkelejda Kasneci

Passwords remain the primary authentication method, yet user-created passwords are often the weakest due to the security-usability trade-off. Although AI-based password generators are emerging, little is known about their effectiveness and user perceptions. This eye-tracking study examined how behavior during password creation, selection, and memorization relates to objective and subjective password quality. Four password models, three AI-based (DeepSeek-API, ChatGPT-API, PassGPT) and one rule-based random generator, generated suggestions from participants' self-generated passwords across four website contexts. Eye movements were recorded throughout the experiment. Results confirm the expected trade-off between AI-generated password strength and human memorability but also reveal a novel behavioral link. Despite stronger AI-generated passwords, participants favored self-generated ones. Notably, visual attention to contextual cues was significantly correlated with higher password entropy. This suggests that security is shaped not only by the generation tool but also by users' visual engagement with contextual cues, highlighting the potential of attention-driven security design.

Mengdi Wang, Anna Bodonhelyi, Efe Bozkir et al.

Federated learning is a distributed collaborative machine learning paradigm that has gained strong momentum in recent years. In federated learning, a central server periodically coordinates models with clients and aggregates the models trained locally by clients without necessitating access to local data. Despite its potential, the implementation of federated learning continues to encounter several challenges, predominantly the slow convergence that is largely due to data heterogeneity. The slow convergence becomes particularly problematic in cross-device federated learning scenarios where clients may be strongly limited by computing power and storage space, and hence counteracting methods that induce additional computation or memory cost on the client side such as auxiliary objective terms and larger training iterations can be impractical. In this paper, we propose a novel federated aggregation strategy, TurboSVM-FL, that poses no additional computation burden on the client side and can significantly accelerate convergence for federated classification task, especially when clients are "lazy" and train their models solely for few epochs for next global aggregation. TurboSVM-FL extensively utilizes support vector machine to conduct selective aggregation and max-margin spread-out regularization on class embeddings. We evaluate TurboSVM-FL on multiple datasets including FEMNIST, CelebA, and Shakespeare using user-independent validation with non-iid data distribution. Our results show that TurboSVM-FL can significantly outperform existing popular algorithms on convergence rate and reduce communication rounds while delivering better test metrics including accuracy, F1 score, and MCC.