News & Research

arXiv:2605.16409v2 Announce Type: replace-cross Abstract: Optical character recognition (OCR) and multilingual text understanding remain major failure modes of multimodal large language models (MLLMs), particularly in real-world images containing cluttered layouts, small fonts, blur, occlusion, and complex typography. We present an OCR-aware multilingual multimodal training framework that combines (i) large-scale synthetic OCR-to-translation data generation, (ii) OCR-aware supervised fine-tuning (SFT) with LoRA adaptation, and (iii) structured visual chain-of-thought (CoT) prompting for reasoning under uncertain visual conditions. Using a LLaMA-based multimodal architecture, the proposed framework substantially improves OCR completeness, multilingual translation accuracy, and robustness under degraded visual conditions. Experimental results on multilingual receipts, menus, posters, signs, handwritten text, and document images demonstrate significantly improved visual-text grounding com

arXiv:2605.16815v2 Announce Type: replace-cross Abstract: Graph neural networks (GNNs) have achieved remarkable success in relational learning. However, their vulnerability to graph backdoor attacks (GBAs) poses a significant barrier to broader adoption in high-stakes applications. Despite recent advances in graph backdoor defense (GBD), existing methods primarily focus on subgraph-based GBAs, relying on the assumption that poisoned target nodes are explicitly connected to subgraph triggers. Our empirical results reveal that such structure-centric approaches fail to defend against emerging feature-based GBAs that preserve graph topology. Therefore, in this paper, we study a novel problem of universal graph backdoor defense. First, we investigate the shared effects of both attack types from a feature-based homophily perspective, which characterizes local feature consistency between nodes and their neighborhoods. Thorough theoretical and empirical analyses demonstrate that, regardless of

arXiv:2605.17788v2 Announce Type: replace-cross Abstract: A fundamental challenge in recommender systems is balancing reliability for Low-Active Users (LAUs) with diversity for High-Active Users (HAUs). The key to this balance lies in quantifying model uncertainty, which approximates the risk of prediction errors and reveals the limits of the model's current knowledge. On large-scale short-video and livestream platforms, model uncertainty can warn of low-quality recommendations that may lead to disengagement of LAUs and at the same time identify opportunities to diversify content recommendation for HAUs. To leverage this dichotomy, we introduce a unified, production-ready framework that calibrates uncertainty to drive differentiated strategies. Specifically, we implement a model-uncertainty-based risk-averse deboosting policy for LAUs to suppress unreliable recommendations, while employing a risk-seeking Upper Confidence Bound (UCB) strategy for HAUs to encourage exploration. Validated

arXiv:2605.18745v2 Announce Type: replace-cross Abstract: Data assimilation (DA) addresses the problem of sequentially estimating the state of a dynamical system from noisy and incomplete observations. In this work, we employ a diffusion model as a world model to simulate and predict the system's dynamics. Recently, score-based diffusion models have learned global diffusion priors that effectively model (stochastic) dynamics, revealing strong potential for data assimilation. In this paper, we investigate how information from noisy observations can be incorporated to enable continuous correction and refinement of the predicted system state when using a diffusion prior. Motivated by particle filtering methods, we represent the posterior distribution using a set of particles. After receiving noisy observations, the diffusion model is guided using the observation likelihood to steer the generation process toward observation-consistent states. Nevertheless, such guidance does not guarantee

arXiv:2605.19170v2 Announce Type: replace-cross Abstract: Diffusion/score-based models have emerged as powerful generative models, capable of generating high-quality samples that mimic the training data distribution. However, it has been observed that they are prone to reproducing training samples-known as "memorization"-potentially violating copyright and privacy. In this paper, we study the effect of Higher-Order Langevin Dynamics (HOLD) on this phenomenon. HOLD diffusion processes introduce auxiliary variables; if the data variable is interpreted as "position," then the auxiliary variables can be interpreted as "velocity" and "acceleration," depending on the chosen order of the model. They were originally proposed based on the intuition that they regularize the trajectories of the data variable by implicitly imposing additional dynamical constraints. Our work provides, to our knowledge, the first theoretical characterization of the regularization effect of HOLD. Specifically, we sho

arXiv:2605.19938v2 Announce Type: replace-cross Abstract: Uniform sampling on implicitly defined manifolds is a core primitive in motion planning, constrained simulation, and probabilistic machine learning. MASEM addresses this problem by entropy-maximizing resampling, but its resampling weights depend on a local k-nearest-neighbour density estimate whose errors can be amplified by aggressive resampling temperatures. We ask whether a polynomial-maximization moment estimator can replace the plug-in density rule without changing the surrounding MASEM architecture. The proposed PMM-MASEM module computes shell spacings from nested k-nearest-neighbour radii, estimates their standardized cumulants, and uses a gated PMM2/PMM3 estimator only when the spacing distribution departs from the flat Exp(1) regime; otherwise it falls back to the plug-in/MLE rule. This fallback is essential: on a flat homogeneous manifold the plug-in estimator is already the MLE, so PMM should not outperform it. A loca

arXiv:2605.20747v2 Announce Type: replace-cross Abstract: Long non-coding RNAs (lncRNAs) are emerging regulatory molecules implicated in chronic disease pathogenesis, including Type 2 Diabetes Mellitus (T2D). We investigated ten literature reported lncRNAs associated with T2D: MALAT1, MEG3, MIAT, ANRIL, GAS5, KCNQ1OT1, H19, BCYRN1, XIST, and HOTAIR across two independent population-based RNA-seq cohorts. Single-omics approaches provide an incomplete view of disease biology, therefore, an integrative multi-feature framework was developed, extracting expression, secondary-structure, and sequence features for each lncRNA. Eight machine learning (ML) classifiers were evaluated under stratified k-fold, leave-one-out cross-validation (LOOCV), and repeated hold-out schemes to ensure robust performance estimation. SHAP analysis was applied for subject-level association interpretation. In one cohort, GAS5 and XIST expression features, along with GAS5, MEG3, and ANRIL sequence features, were fou

arXiv:2605.24142v1 Announce Type: new Abstract: Metacognitive theories provide foundational frameworks for understanding self-regulated learning, yet they lack systematic integration into comprehensive scenario taxonomies capable of guiding AI-enhanced professional development interventions. Existing models inadequately specify how metacognitive components combine into distinct learning scenarios or how professionals progress from novice to expert functioning. A six-node open systems model, consisting of Environment, Input, Processes, Structures, Output, and Feedback, was developed by synthesizing four major theoretical frameworks. Combinatorial enumeration generated 216 mathematically possible learning scenarios. Four sequential constraint-based filters, including psychological plausibility, educational relevance, measurement feasibility, and intervention potential, informed by empirical workplace learning research, reduced this space to 24 priority scenarios. Five focal scenarios wer

arXiv:2605.24228v1 Announce Type: new Abstract: We investigate sketch-like pen input as an alternative way to support execution control in interactive debugging. In our interface, programmers draw lightweight marks to set breakpoints, use symbolic strokes to control execution, and extend strokes into spirals to repeat traversal actions. The prototype combines gesture recognition with Python execution tracing in a conventional editor interface. In a controlled study with 24 programmers, we compared the sketch interface with conventional mouse-and-keyboard input on debugging tasks that required breakpoint placement, step-wise execution, and runtime state inspection. The results show that sketch-like input can support these execution-control tasks, while also introducing challenges in precision, recognition, and gesture recall. Our findings suggest that pen input is most promising where debugger interactions benefit from spatial grounding or continuous movement, rather than as a wholesale

arXiv:2605.24307v1 Announce Type: new Abstract: Privacy measurement instruments (e.g., CFIP, IUIPC, PAQ) predate GDPR by over a decade and measure privacy concerns, distinct from preferences for regulatory protections (e.g., data portability, erasure, automated decision-making rights). This leaves practitioners without tools to assess whether users value the GDPR mechanisms implemented in compliant policies. We developed a GDPR-grounded privacy preference measurement item bank by extracting 669 statements from all 99 GDPR articles, validated by: (1) two-round expert review achieving full consensus on accuracy, (2) semantic clustering into 10 parent themes and 87 subthemes, and (3) consensus review with 50 privacy experts (5 per theme) using a larger or equal than 4/5 vote retention threshold. The final 527-item bank comprises 9 parent themes and 73 subthemes (18 to 112 items per parent theme, 1 to 29 per subtheme), enabling targeted measurement across granularities while covering GDPR

arXiv:2605.24332v1 Announce Type: new Abstract: Current generative AI systems are increasingly effective at processing explicit knowledge, including retrieving information, summarising documents, generating explanations, and supporting codified workflows. However, high-level expertise also depends on tacit sensing: perceiving weak signals, recognising emerging tensions, detecting coherence degradation, and anticipating instability before formal indicators appear. Existing AI education, AI literacy, and human-AI collaboration frameworks remain centred on prompting, task execution, and productivity support and are poorly equipped to address this tacit layer of expert cognition. This vision paper argues that next-generation AI systems should move beyond explicit knowledge processing toward the longitudinal modelling of expert tacit sensing. It introduces Tacit Signal Infrastructure as a layer for capturing, structuring, modelling, interpreting, and validating expert tacit signals over tim

arXiv:2605.24337v1 Announce Type: new Abstract: Voice is a central element of identity. We recognize people by their voice, and we uniquely express who we are with it. For people who rely on augmentative and alternative communication~(AAC) systems, such as speech-generating devices~(SGD), the device's voice becomes an identity marker others associate with them. Yet, it is hard to find a voice that truly aligns with one's identity both linguistically and culturally. Although modern AI-generated voices can reproduce diverse accents and speaking styles, AAC users still lack accessible ways to articulate how they want an identity-aligned voice to sound like. We first conducted a survey of AAC users (across eight countries) to characterize current voice representation, finding that non-binary, transgender, and non-US-born respondents rated their current voice support identity alignment consistently lower than other respondents. To examine how AAC users respond to voices designed to reflect

arXiv:2605.24729v1 Announce Type: new Abstract: While human-AI collaboration systems have increasingly been built to increase efficiency or support creativity, little work has examined how the design of interactions shapes the social connection between human and artificial agent. We examine how the temporal and visual dimensions of collaboration shape the experience of a writing task. Specifically, we built three variants of an AI-assisted text editor along a spectrum of simulated humanlike interaction (synchronous and with a cursor) to machinelike interaction (asynchronous and without a cursor), and conducted a comparative user study (n=48). Our exploratory findings suggest that synchronous suggestions increased efficiency but led to contextual misalignment, while a visual cursor increased intent understanding but evoked feelings of surveillance. Taken together, humanlike design of artificial agents can create positive social expectations but also elicit social costs, especially witho

arXiv:2605.24830v1 Announce Type: new Abstract: As personal agents evolve to handle complex, user-centric tasks, static plain-text chat is rapidly becoming a bottleneck. Generative UI emerges as the necessary new interface layer, dynamically synthesizing the right controls, options, and state from the interaction context in real time. We present Macaron-A2UI, a model for Generative UI in personal agents. Our goal is to move beyond text-only interaction by enabling agents to generate natural language together with lightweight, executable UI actions for information collection, preference refinement, confirmation, and multi-goal organization. We build a large-scale Generative UI corpus from heterogeneous dialogue sources, introduce A2UI-Bench for controlled evaluation, and train 30B, 235B and 754B models with parameter-efficient LoRA-based supervised fine-tuning followed by reward-driven reinforcement learning. The best Macaron-A2UI model reaches 75.6 overall on A2UI-Bench without explici

arXiv:2605.25260v1 Announce Type: new Abstract: This paper offers a concept of working relations as a complement and extension to existing theories of maintenance, care and repair. Building on the cases of an umbrella, a tractor and a pond, it advances seven propositions that might guide and inform further work and thinking in this space. It concludes with the challenging figures of Chernobyl, nickel extraction, and AI, and argues for the centrality of working relations to more generative and pluralistic relations with the things and worlds around us.