arXiv:2606.25514v1 Announce Type: new Abstract: Resolving issues with ambiguous and incomplete descriptions, particularly concerning complex bugs, requires a sophisticated, long-horizon workflow. Agents must navigate codebases to locate the root cause, reproduce the failure, implement a fix, and validate the resulting patch. Inefficient context management, thereby, can lead to rapid context degradation and context poisoning, preventing successful resolution. We propose icat-agent, a decentralized, multi-agent scaffolding that replaces shared context with synchronous, event-based message passing. Utilizing a rubric-based issue quality check, icat-agent strategically pivots its workflow: it initiates parallel patching and validation for well-defined issues, while deploying preliminary exploration for low-quality ones. A comprehensive evaluation of icat-agent on SWE-bench Verified and SWE-bench Pro demonstrates that it consistently outperforms prominent baselines across all difficulty levels, including SWE-agent, mini-SWE-agent, and Claude Code, while using the same underlying models, improving by 3.6-8.4% on SWE-bench Verified and 6.3-18.5% on SWE-bench Pro. icat-agent is also computationally efficient, reducing the average cost by $1.18 per instance compared with the multi-agent Claude Code baseline. Our findings reveal that a robust scaffold such as icat-agent unlocks substantial latent capability within a fixed model, with the same backbone resolving markedly more issues under icat-agent than under existing scaffolds. icat-agent +GPT-5.4-xhigh resolves 67.4% of SWE-bench Pro problems, outperforming the current best result on SWE-bench Pro (59.10%, mini-SWE-agent+GPT-5.4-xhigh) by 8.3 percentage points.
Science Journals
arXiv:2606.25520v1 Announce Type: new Abstract: In this work, we consider a two-dimensional time-space fractional diffusion equation with a variable coefficient and investigate the inverse source problem of reconstructing the source term f(x,y) , after regularizing the problem using the quasi-boundary value method to mitigate ill-posedness. A finite difference discretization results in a large-scale linear system with a multilevel Toeplitz-like block structure. We perform a spectral analysis of the associated matrix sequences, employing tools from Generalized Locally Toeplitz (GLT) theory, and construct efficient preconditioners based on the GLT analysis. The proposed preconditioners preserve the multilevel structure of the discretization matrices and leads to a general eigenvalue clustering around one for the preconditioned sequence. Numerical experiments validate the theoretical findings and demonstrate that the proposed approach significantly accelerates the convergence of the GMRES method in reconstructing the source term in the two-dimensional space-time fractional diffusion equation.
arXiv:2606.24961v1 Announce Type: cross Abstract: We study the strict locked-orientation limit of hard spherocylinders on a sphere, in which the rod axes are rigidly locked to a prescribed tangential director field and cannot reorient. Because the bulk hard-rod phase diagram contains no smectic-C phase, any coherent tilt isolates a geometric curvature mechanism rather than a finite-stiffness equilibrium effect. A ratio-symmetric recognition cost fixes the layer spacing at the bulk close-contact value and yields a hierarchy of geometric statements: the lower edge of the smectic-area window at $45^\circ$ follows from reciprocal symmetry; the upper edge at $58.3^\circ$ is a falsifiable channel-saturation hypothesis; the smectic-A to smectic-C boundary is a closed-form prediction; and the rod tilt angle is set by the rod-to-radius ratio, modulated by a chirality envelope peaking near $24^\circ$. Locked-orientation Monte Carlo across fifteen geometries confirms these predictions with no fitted elastic constants: the smectic area peaks at $55^\circ$, and a coherent smectic-C window is detected.
arXiv:2606.25254v1 Announce Type: cross Abstract: Maternal-fetal US is the primary imaging modality for monitoring fetal development, yet accurate automated segmentation remains challenging due to the scarcity of pixel-level annotations. To address this issue, we propose DACL, a semi-supervised framework for robust fetal US image segmentation. DACL jointly trains a deployment-oriented lightweight convolutional network (1.47\thinsp\mathrm{M} parameters) and a Transformer-based network, leveraging labeled data for supervised learning and unlabeled data via CPS. To enhance prediction stability, we introduce a dual-agreement consistency loss that couples pixel-wise probabilistic divergence with entropy-guided confidence alignment. Unlike conventional CPS methods that enforce agreement only at the prediction level, DACL explicitly regularizes both distributional alignment and uncertainty, thereby suppressing unreliable pseudo-labels and enabling stable cross-architecture pseudo-label learning under extreme annotation scarcity. Furthermore, an interpolation-based consistency strategy using mixup is applied to unlabeled samples to enhance robustness. Under 5% labeled data, DACL improves Dice by up to 2.77% and reduces HD95 by up to 14.69 mm compared with the strongest recent semi-supervised methods, demonstrating significant improvements in boundary accuracy on both fetal head and abdomen datasets. These results demonstrate the effectiveness of agreement-based consistency learning for annotation-efficient fetal US segmentation. Our code is on GitHub.
arXiv:2606.25267v1 Announce Type: cross Abstract: Rapid and robust laser-frequency auto-locking is essential for the field deployment of quantum communications, quantum computing, and precision-measurement technologies; however, achieving this remains a considerable challenge. Here, we propose and demonstrate an auto-locking scheme employing Bayesian optimization and discrete biorthogonal wavelet transformation. First, the reference is rapidly sought by making intelligent use of historical observations, eliminating the inherent blindness of the traditional parameter-scanning method. Second, the frequency reference is robustly identified by pinpointing transition signals with the discrete biorthogonal wavelet transformation and analyzing their immutable frequency differences and relative magnitudes, which are determined by the inherent atomic structure and remain resistant to environmental disturbances. This proposed approach achieves a fivefold acceleration in reference searching compared to conventional scanning methods in the case where the laser frequency drifts far away from the reference. Crucially, it achieves an identification accuracy of more than 99.5 %, even under severe 50 % laser-intensity fluctuations, $9.95^\circ$ photodiode misalignment, and $18^\circ$C Rb cell temperature elevation. Finally, locking the laser frequency to the identified reference with a lead zirconate titanate-current double-servo loop narrows the linewidth to 20 kHz. We believe that this rapid, robust, and high-performance auto-locking technique will be pivotal towards the deployment of the next generation of practical quantum technologies in demanding field environments.
arXiv:2606.25269v1 Announce Type: cross Abstract: As black-box models become foundational to modern research, ensuring their stability is paramount for the realization of trustworthy artificial intelligence. The inherent diversity of inputs - ranging from structured Gaussian distributions to complex data with unknown structures - poses a significant challenge: how to stabilize black-box outputs while effectively leveraging available prior information. This paper introduces a task-oriented randomization methodology that adaptively tailors its strategy to the underlying generative mechanisms of the input data, specifically addressing unstructured complexities. A comprehensive suite of stability guarantees is proposed. Beyond establishing rigorous theoretical foundations for stability, the research provides a detailed analysis of the intrinsic trade-off between stability and exploration. Motivated by the architecture of Large Language Models, the framework is further extended to top-k ranking problems. The validity and effectiveness of the proposal are demonstrated through extensive numerical simulations and applications to the real-world dataset.
arXiv:2606.25012v1 Announce Type: new Abstract: Many reinforcement learning (RL) problems in the infinite-horizon average-reward setting require optimizing multiple conflicting objectives while satisfying multiple safety constraints. A common approach is concave scalarization, where the agent maximizes a utility $ f(J^\pi_{r_1}, \ldots, J^\pi_{r_M}) $ subject to a scalarized constraint $ g(J^\pi_{c_1}, \ldots, J^\pi_{c_N}) \ge 0 $, where $J^\pi_{r_m}$ and $J^\pi_{c_n}$ denote the average-reward and cost under policy $\pi$. However, the nonlinearity of $f$ and $g$ introduces bias in policy-gradient and actor-critic methods, since gradients must be evaluated using noisy estimates of $J^\pi,$ and $ \mathbb{E}[\partial f(J^\pi)] \neq \partial f(\mathbb{E}[J^\pi]),$ and this bias propagates through both primal and dual updates. We propose an MLMC-based primal-dual Natural Actor-Critic algorithm for average-reward MDPs that controls bias in scalarized objectives, constraint evaluation, and actor-critic estimation without requiring mixing-time knowledge. We show that the algorithm achieves optimal global convergence and constraint-violation rates of $ \tilde{O}(1/\sqrt{T}) $. To our knowledge, this is the first result establishing optimal convergence for concave scalarized multi-objective RL in the average-reward setting, both with and without constraints, and the first to do so without mixing-time information even in the absence of scalarization.
arXiv:2606.25039v1 Announce Type: new Abstract: Recovering governing Ordinary Differential Equations (ODEs) from data is a central challenge in modeling dynamical systems across scientific domains. Existing approaches cast discovery as a static inference problem over fixed datasets, assuming that the observed trajectories are sufficiently informative. However, dynamical systems evolve over large state spaces, and limited data can make multiple equations observationally indistinguishable, leading to identifiability gaps and the recovery of incorrect governing equations. To address this, we introduce LLM-ACES, or LLM-guided Active Closed-loop Equation Search, a closed-loop framework that jointly optimizes symbolic hypothesis construction and adaptive data acquisition. In LLM-ACES, a large language model (LLM) proposes operator priors that partition the large search space into distinct regions, within which candidate equations are fit to the observed data. The disagreement among these candidates guides the acquisition of informative trajectories, creating a feedback loop that iteratively refines both the hypothesis space and the discovered dynamics. On 122 ODE systems spanning ODEBench and ODEBase, LLM-ACES achieves the lowest median NMSE, outperforming state-of-the-art baselines by several orders of magnitude while achieving a high symbolic accuracy of 46.2% and 52.4%, respectively. Our analysis further shows that LLM-ACES is sample-efficient, achieving better performance with one-tenth the data. Furthermore, LLM-ACES's feedback-driven data acquisition makes it robust to noise and recovers the correct symbolic structure, while baselines introduce spurious terms that fit the data locally but obscure the true governing relationships.
arXiv:2606.25557v1 Announce Type: cross Abstract: The isoreticular index of the eight-member embedded RHO zeolite hierarchy operates as a phenomenological design knob that tunes the mechanical critical pressure of the framework valve from $\sim 0.94$ GPa for parent RHO down to a predicted $\lesssim 0.03$ GPa for the largest member PST-28, more than an order of magnitude lower across a single family of synthesizable nanoporous solids. The molecular-valve effect that defines zeolite RHO (a reversible centric-to-acentric phase transition triggered by water, gas pressure, or mechanical loading) is shown here to be not a peculiarity of the smallest member but a generic property of the whole hierarchy, with a critical pressure that decays exponentially with the isoreticular order $k$. Combining lattice dynamics, full elastic-constant tensors, and finite-temperature free-energy reconstructions within a classical core-shell description of the pure-silica frameworks (independently validated against r$^2$SCAN+rVV10 density-functional theory on $G_1$ and $G_2$), we find that the entire family is well described by an effective mean-field Landau picture in which the framework distortion couples quadratically to the volumetric strain. We emphasise that $p_c$ is a mechanical (hydrostatic) critical pressure of the bare pure-silica framework, used as a proxy for the intrinsic framework softness and for the cation- and dehydration-driven response of the real aluminosilicates; it is not a gas-adsorption pressure. The exponential extrapolation places $p_c(G_6\text{-}G_8) \lesssim 0.1$ GPa (model-dependent band $0.03$-$0.15$ GPa), identifying the higher-order members as the softest, most stimuli-responsive frameworks of the hierarchy; whether this intrinsic softness translates into guest-driven switching at low gas activities will depend on the Al distribution, extra-framework cations and adsorbed molecules, and remains to be tested experimentally.
arXiv:2606.25615v1 Announce Type: cross Abstract: In this paper, we consider a parabolic problem with constraint written as a differential inclusion, driven by a multiplicative colored noise and involving a p-Laplace operator (for $p \geq 2$), nonlinear random source terms and subject to Neumann boundary conditions on a bounded Lipschitz domain of $R^d$ with $d \geq 1$. This contribution aims at proving existence and uniqueness of a solution for such a multivalued problem. On one hand, the existence result is proved by the analysis of a semi-implicit time discretization scheme constructed on a smoother version of our problem, itself obtained by a regularization "\`a la Moreau-Yosida" of the subdifferential term. The key point of our approach consists in finding a clever relation between the time step denoted $\tau$ and the Moreau-Yosida regularization parameter denoted $\epsilon$ in view to pass simultaneously to the limit with respect to $\tau$ and $\epsilon$. On the other hand, the uniqueness of the solution is proved by standard arguments.
arXiv:2606.25945v1 Announce Type: cross Abstract: The increasing number of orbital and sub-orbital launches makes it necessary to investigate the environmental impacts of launch vehicles and incorporate eco-design considerations into their development. In response, the European Space Agency has promoted Life Cycle Assessment (LCA) as a standardization methodology to mitigate environmental impacts of present and future space missions. This need is further amplified in the NewSpace, where numerous configurations and innovative technologies are explored, reinforcing the importance of integrating environmental considerations. At early design stages, launch vehicle architecture can be formalized through a multi-physics optimization problem based on Multidisciplinary Design Analysis and Optimization (MDAO) methods, where disciplines such as propulsion, aerodynamics, structure, and trajectory are coupled to obtain trade-offs among candidate configurations. This paper proposes a methodology to integrate an LCA discipline within an MDAO framework for launch vehicle design. The approach relies on parametric life-cycle inventories depending on design and coupling variables, covering component and propellant production as well as transport to the launch site. Launch emissions are evaluated from optimized trajectory profiles and characterized in terms of climate change impact. The methodology is illustrated on a representative expendable launch vehicle, where multi-objective optimizations assess trade-offs between performance and environmental indicators. Results highlight antagonistic behaviors among environmental impact categories, emphasizing the importance of carefully defining environmental objectives in eco-design studies. The generic nature of the methodology lays the foundation for integrating LCA into early-stage launch vehicle design, enabling exploration of trade-offs between performance, cost, and environmental considerations.
arXiv:2606.25626v1 Announce Type: new Abstract: We present a general answer set programming based hybrid quantitative-qualitative method for computing constrained branching trajectory modes for moving objects in real-world settings. The method performs constrained traversal of an environment graph, enumerating geometrically admissible motion behaviours as stable models, each constituting a distinct trajectory mode characterised by both domain-dependent and independent factors such as derived event sequence, map topology, and domain norms. The hybrid trajectory computation method is generally applicable across motion characteristics typically encountered in diverse dynamic domains with moving objects, e.g., autonomous driving. We demonstrate applicability and highlight how computed trajectories are traceable to their underlying stable model, thereby affording verifiable interpretability that purely learned approaches cannot provide. We also perform an empirical evaluation with Argoverse 2, a large-scale real-world autonomous driving benchmark representative of the class of dynamic domains within the scope of the proposed method.
arXiv:2507.20068v3 Announce Type: replace Abstract: Off-policy evaluation (OPE) methods estimate the value of a new reinforcement learning (RL) policy prior to deployment. Recent advances have shown that leveraging auxiliary datasets, such as those synthesized by generative models, can improve the accuracy of OPE methods. Unfortunately, such auxiliary datasets may also be biased, and existing methods for using data augmentation within OPE lack principled uncertainty quantification. In high stakes domains like healthcare, reliable uncertainty estimates are important for ensuring safe and informed deployment of RL policies. In this work, we propose two methods to construct valid confidence intervals for OPE with data augmentation. The first provides a confidence interval over $V^{\pi}(s)$, the policy value conditioned on an initial state $s$. To do so we introduce a new conformal prediction method suitable for Markov Decision Processes (MDPs) with continuous state spaces, extending prior work to higher-dimensional settings. Second, we consider the more common task of estimating the average policy performance over all initial states, $V^{\pi}$; we introduce a method that draws on ideas from doubly robust estimation and prediction powered inference. Across simulators spanning inventory management, robotics, healthcare, and a real healthcare dataset from MIMIC-IV, we find that our methods can effectively leverage auxiliary data and consistently produce confidence intervals that cover the ground truth policy values, unlike previously proposed methods. Our work enables a future in which OPE can provide rigorous uncertainty estimates for high-stakes domains.
arXiv:2606.25658v1 Announce Type: new Abstract: Currently, streaming video understanding is still a daunting task for existing \emph{multimodal large language models} (MLLMs). Its difficulties not only lie in handling the ever-increasing video frames, but also in the unpredictability of future video content and input instructions. In this paper, we study this task from the perspective of constructing a dynamic but fixed-budget memory bank, and propose a novel and training-free approach termed \emph{\textbf{CausalMem}}. CausalMem is dedicated to constructing a dynamic visual memory update mechanism, thereby maximizing the amount of information in streaming video within a limited memory space, much like the human brain. In practice, CausalMem estimates the redundancy of visual tokens and updates the memory bank via an online semantic basis, which models the principal semantics of the observed video stream. To validate CausalMem, we apply it to two representative MLLMs, namely LLaVA-OneVision and Qwen2.5-VL respectively, and conduct extensive experiments on both streaming and offline video understanding benchmarks. The experimental results not only show the great advantages than existing methods under both streaming and offline settings, \emph{e.g.}, $+3.2\%$ and $+3.0\%$ average accuracy gains respectively, but also witness the superior semantic preservation for streaming videos, \emph{e.g.}, using 12$k$ token budgets to memorize hour-long streaming videos, which achieves more than \textbf{20$\times$} visual token compression ratio and only occupies about \textbf{82 MB} storage. \textbf{Our code} is given in \href{https://github.com/hktk07/CausalMem}{CausalMem}.
arXiv:2606.25659v1 Announce Type: new Abstract: Learning-based Lyapunov Control (LLC) provides formal stability guarantees for nonlinear systems, but its validity relies heavily on accurate system models. Parameter variations and uncertainties may invalidate stability constraints, leading to costly retraining. Although D-learning can estimate Lyapunov derivatives without relying on explicit dynamics models, it remains limited by single-task dynamics and degrades under large parameter shifts. We propose Reptile-D-learning, a framework that leverages the Reptile meta-learning algorithm to capture shared dynamical structures across systems with different parameters, thereby learning a generalizable Lyapunov network initialization and a high-performance controller. Experiments on multiple nonlinear control systems demonstrate that Reptile-D-learning significantly improves both generalization and rapid adaptation to unseen parameter configurations.
arXiv:2509.13584v2 Announce Type: replace Abstract: The k-core of a graph is its maximal subgraph with minimum degree at least k, and the core value of a vertex u is the largest k for which u is contained in the k-core of the graph. Among cohesive subgraphs, k-core and its variants have received a lot of attention recently, particularly on dynamic graphs, as reported by Hanauer, Henzinger, and Schulz in their recent survey on dynamic graph algorithms. We answer questions on k-core stated in the survey, proving that there is no efficient dynamic algorithm for k-core or to find (2 - {\epsilon})-approximations for the core values, unless we can improve decade-long state-of-the-art algorithms in many areas including matrix multiplication and satisfiability, based on the established OMv and SETH conjectures. Some of our results show that there is no dynamic algorithm for k-core asymptotically faster than the trivial ones. This explains why most recent research papers in this area focus not on a generic efficient dynamic algorithm, but on finding a bounded algorithm, which is fast when few core values change per update. However, we also prove that such bounded algorithms do not exist, based on the OMv conjecture. We present lower bounds also for a directed version of the problem, and for the edge variant of the problem, known as k-truss. On the positive side, we present a polylogarithmic dynamic algorithm for 2-core.
arXiv:2606.24899v1 Announce Type: new Abstract: AI-assisted mathematics is often evaluated on solving predefined problems. In practice, however, many important advances begin earlier, when a vague research intuition is transformed into a concrete problem, a promising route, and a theorem family worth proving. This report studies that stage through a case study that led to sign-embedding quantum algorithms for matrix equations and matrix functions, foundational primitives in quantum linear algebra and operator-output quantum algorithms. The project began with a human-originated intuition that rational approximation is especially effective for jump-type functions such as the sign function, and might therefore serve as a design principle for quantum algorithms. Rather than merely assisting after the problem was fixed, AI-assisted exploration, including workflows later integrated into the agentic AI-mathematician system AIM, played a key role in expanding this intuition into a route map, comparing candidate formulations, and converging toward sign embedding as the central framework. AIM then helped connect a known matrix-sign identity to wider classes of matrix equations and matrix functions, and drafted proof and complexity calculations. The decisive scientific judgments remained human: selecting which human-AI-expanded routes were worth pursuing, rejecting a Cayley-trapezoidal approximation when its validity required a hidden condition, and refining the Sylvester implementation from a coarse quadratic-gap query route to the final factorized and scaled analysis. The report argues that human-AI co-discovery workflows, with systems such as AIM as important components, are most valuable not as standalone theorem provers, but as research partners for problem formation, connection discovery, derivation, and skeptical review inside a human-gated research loop.
arXiv:2606.25681v1 Announce Type: new Abstract: Motion-to-photon latency is inherent in immersive virtual reality (VR) systems and can arise from multiple sensorimotor loops, including view-contingent latency between head movement and display update and interaction latency between hand movement and the virtual effector. Although prior work shows that interaction latency can impair VR performance, it remains unclear whether common spatial, temporal, and efficiency measures reveal the same latency-related disruption. This study addressed this question by experimentally imposing delays between the physical and virtual hands during manual pointing in VR. Participants pointed to targets on a horizontal surface in VR and in the physical environment as an unmediated baseline. In VR, pointing was performed with a virtual hand avatar controlled by a motion capture pipeline, and additional delays (0-500 ms) were imposed between the participant's hand movement and the rendered movement of the virtual hand. Relative to the baseline, performance in VR showed greater endpoint error, longer movement time, greater endpoint variability, and lower throughput. Within VR, added interaction latency further increased endpoint error and variability, reduced throughput, and altered movement time, but these effects followed different profiles: endpoint error increased even at the shortest delays, whereas movement time remained stable at short delays and increased primarily at longer delays. These findings show that interaction latency produces dissociable spatial and temporal consequences in immersive VR, such that endpoint accuracy revealed disruption before movement time or throughput. Thus, latency-sensitive VR interactions cannot be fully evaluated using movement time or efficiency measures alone. Instead, HCI evaluations should assess both spatial and temporal performance, particularly when VR tasks involve visually guided manual actions.
arXiv:2509.19297v3 Announce Type: replace Abstract: Feed-forward 3D Gaussian Splatting (3DGS) has emerged as a highly effective solution for novel view synthesis. Existing methods predominantly rely on a \emph{pixel-aligned} Gaussian prediction paradigm, where each 2D pixel is mapped to a 3D Gaussian. We rethink this widely adopted formulation and identify several inherent limitations: it renders the reconstructed 3D models heavily dependent on the number of input views, leads to view-biased density distributions, and introduces alignment errors, particularly when source views contain occlusions or low texture. To address these challenges, we introduce VolSplat, a new multi-view feed-forward paradigm that replaces pixel alignment with voxel-aligned Gaussians. By directly predicting Gaussians from a predicted 3D voxel grid, it overcomes pixel alignment's reliance on error-prone 2D feature matching, ensuring robust multi-view consistency. Furthermore, it enables adaptive control over density based on 3D scene complexity, yielding more faithful Gaussians, improved geometric consistency, and enhanced novel-view rendering quality. Experiments on widely used benchmarks demonstrate that VolSplat achieves state-of-the-art performance, while producing more plausible and view-consistent results. The video results, code and trained models are available on our project page: https://lhmd.top/volsplat.
arXiv:2509.22193v2 Announce Type: replace Abstract: Distilling reasoning traces from strong teacher models has become the standard recipe for building capable small language models. Yet reasoning traces are 5-20$\times$ longer than standard instruction fine-tuning (IFT) outputs, meaning every practitioner who chooses reasoning distillation implicitly forgoes training a larger IFT model on the same compute budget. Whether this trade-off is worthwhile remains unaddressed. We study it with a controlled experiment: a single teacher generates paired IFT and reasoning outputs for identical prompts by toggling only its reasoning mode, isolating supervision format as the sole variable. Training students at five scales (0.5B to 14B) and evaluating on 18 benchmarks, we find that at matched FLOPs, IFT lies on or near the Pareto frontier across the majority of configurations. Reasoning reaches the Pareto frontier only on open-ended tasks at 7B and above. Even there, a sequential curriculum mixing just 25-50\% reasoning data with IFT captures most of the accuracy benefit at far lower compute cost.
arXiv:2606.25719v1 Announce Type: new Abstract: In this paper, we introduce position graphs, a graph-based reasoning framework based on the formalization of position spaces. This framework utilizes two strict partial orders, representing horizontal and vertical alignment and precedence, to model the relative positions of discrete tokens. Unlike general qualitative spatial calculi, position graphs are constrained by a chain condition and compatibility requirements that focus on rows and columns. We provide a comprehensive theoretical analysis of this representation, beginning with a characterization of graph consistency. Conditions to ensure the consistency of position graphs are established. Furthermore, we investigate the computational complexity of structural pattern discovery, modeled as the induced subgraph isomorphism problem. We demonstrate that this problem remains NP-complete even within the restricted class of position graphs. While initially motivated by document processing, this work focuses on the underlying mathematical properties and algebraic consistency of position-based constraints, providing a formal logical layer that is independent of specific data extraction techniques.
arXiv:2509.26376v3 Announce Type: replace Abstract: Test-time strategies have shown remarkable success in improving large language models, but their application to next-token prediction (NTP) autoregressive (AR) image generation remains largely underexplored. Existing test-time scaling (TTS) methods for visual autoregressive models (VAR) rely on frequent partial decoding and external reward models, which are inefficient and often ineffective for NTP-based image generation due to the inherent instability of intermediate decoding results. To address these limitations, we propose ScalingAR, a novel test-time scaling framework tailored for NTP-based AR image generation. ScalingAR introduces token entropy as a confidence signal and operates at two complementary levels: (i) Profile Level, integrates intrinsic uncertainty and conditional utilization into a unified confidence state, and (ii) Policy Level, leverages this state for adaptive trajectory pruning and dynamic guidance scheduling. Without requiring early decoding or auxiliary rewards, ScalingAR achieves significant improvements across diverse benchmarks. Experiments show that ScalingAR (I) improves base models by $12.5\%$ on GenEval and $15.2\%$ on TIIF-Bench, (II) reduces visual token consumption by $62.0\%$ while outperforming baselines, and (III) enhances robustness, mitigating performance degradation by $26.0\%$ in challenging scenarios. These results establish ScalingAR as a robust and efficient test-time scaling solution for autoregressive image generation.
arXiv:2512.00804v3 Announce Type: replace Abstract: When answering user queries, LLMs often retrieve knowledge from external sources stored in retrieval-augmented generation (RAG) databases. These are often populated from unvetted sources, e.g. the open web, and can contain maliciously crafted data. This paper studies attacks that can manipulate the context retrieved by LLMs from such RAG databases. Prior work on such context manipulation primarily injects false or toxic content, which can often be detected by fact-checking or linguistic analysis. A more subtle threat, which we call epistemic bias injection (EBI), is where adversaries inject factually correct yet epistemically biased passages that systematically favor one side of an open-ended issue. Although linguistically coherent and truthful, such adversarial passages effectively crowd out alternative viewpoints during retrieval from the RAG and push LLM outputs towards an attack-desired stance. As a core contribution, we propose a novel characterization of the problem: We give a geometric metric that quantifies stance polarity and epistemic bias. This metric can be computed directly on embeddings of text passages. Leveraging it, we construct EBI attacks and develop a lightweight prototype defense called BiasDef for them. We evaluate them both on a comprehensive benchmark constructed from public question answering datasets. Our results show that: (1) the proposed attack induces significant stance polarity shifts, effectively evading existing retrieval-based sanitization defenses, and (2) BiasDef substantially reduces adversarial retrieval and epistemic bias in LLM's answers. Overall, this demonstrates the new threat as well as the ease of employing epistemic bias metrics for filtering in RAG-enabled LLMs.
arXiv:2512.24946v3 Announce Type: replace Abstract: Existing film-restoration methods frequently fail under fast motion, producing limb disappearance and structural distortion due to inaccurate motion modeling. Moreover, high-resolution restoration under spatially-persistent and mixed defects remains insufficiently studied. We propose HaineiFRDM, a Film Restoration Diffusion Model that leverages the content modeling capability of diffusion models for content-aware restoration, removing defects while preserving scene structure.To enable scalable high-resolution restoration, we adopt a patch-wise strategy with position-aware global fusion modules to maintain cross-patch coherence. We further introduce a frequency-based module to enhance texture consistency and a patch-consistent inference framework to alleviate blocking artifacts introduced by patch-based processing.We also construct a film restoration dataset comprising categorized defect templates, professionally restored films, and realistic synthetic degradations.Extensive experiments demonstrate our superior restoration quality with strong structural consistency. Our design also reduces memory requirements, enabling high-resolution restoration on a single 24GB-VRAM GPU.Code and the dataset will be released at https://anonymous.4open.science/r/HaineiFRDM.
arXiv:2606.24978v1 Announce Type: new Abstract: Accurate particulate matter (PM) prediction is crucial for mitigating air pollution. Graph Neural Networks (GNNs) effectively model spatiotemporal dependencies, but predefined graphs limit adaptability, and some datasets complicate learning. This study introduces a graph construction method based on a confusion matrix from a supervised learning process to dynamically capture inter-class relationships. Additionally, a hybrid loss function that combines energy distance and Huber loss is applied to address the vanishing gradient problem and improve learning stability. The approach is evaluated using air pollution data from the University of Utah AirU Pollution Monitoring Network in Salt Lake City, UT, with five GNN models: Graph Convolutional Networks (GCNs), Simple Graph Convolutional Networks (SGConv), Graph Isomorphism Networks (GINs), Graph Attention Networks (GATs), and GraphSage. The experimental results of single- and multistep predictions confirm that GraphSage achieves the highest accuracy in predicting the concentrations of PM${1}$, PM${10}$, and PM$_{2.5}$ over different time horizons. Furthermore, {\color{black} GNNExplainer (Graph Neural Network Explainer) and PGExplainer (Probabilistic Graph Explainer)} are applied to interpret feature importance and graph structure, ensuring model transparency. Results show improved prediction accuracy, with GNN models outperforming traditional machine learning \textcolor{black}{and deep learning models (i.e., Prophet, Long short-term memory, Gated recurrent units} in air pollution forecasting.