arXiv:2511.08195v4 Announce Type: replace Abstract: UI-to-code aims to translate UI screenshots into executable front-end code. Despite progress with vision-language models (VLMs), most existing methods formulate UI-to-code as a single-pass generation, which mismatches real-world UI development that is inherently iterative and feedback-driven. We reformulate UI-to-code as an interactive visual optimization problem, where code generation is embedded in a closed-loop process of execution, visual inspection, and iterative refinement driven by rendered visual feedback. To address the non-differentiability of visual objectives and the noise of absolute visual evaluators, we propose Relative Visual Policy Optimization (RVPO), a preference-based reinforcement learning method that optimizes relative visual rankings among rendered candidates under execution feedback. We instantiate this paradigm in UI2Code^N, an open-source 9B model trained via continual pre-training, supervised fine-tuning, and reinforcement learning. Experiments demonstrate state-of-the-art performance on UI drafting, UI polishing, and UI editing benchmarks, even outperforming larger models, with performance consistently improving through iterative visual optimization. Our code and models are available at https://github.com/zai-org/UI2Code_N.
Science Journals
arXiv:2511.08299v2 Announce Type: replace Abstract: Salamander-like robots are designed inspired by the skeletal structure of their biological counterparts. However, existing controllers cannot fully exploit these morphological features and largely rely on predefined patterns or joint trajectories, which prevents the generation of diverse and flexible gaits and limits their applicability in real-world scenarios. In this paper, we propose a phase-based learning framework that enables the robot to acquire a diverse repertoire of gaits without using reference motions. Each body part is controlled by a phase variable capable of forward and backward evolution, with a phase coverage reward to promote the exploration of the leg phase space. Additionally, morphological symmetry of the robot is incorporated via data augmentation, improving sample efficiency and enforcing both motion-level and task-level symmetry in learned behaviors. Extensive experiments show that the robot successfully acquires 22 representative gaits exhibiting both dynamic and symmetric movements, demonstrating the effectiveness of the proposed learning framework.
arXiv:2511.14427v4 Announce Type: replace Abstract: Effective contact-rich manipulation requires robots to synergistically leverage vision, force, and proprioception. However, Reinforcement Learning agents struggle to learn in such multisensory settings, especially amidst sensory noise and dynamic changes. We propose MultiSensory Dynamic Pretraining (MSDP), a novel framework for learning expressive multisensory representations tailored for task-oriented policy learning. MSDP is based on masked autoencoding and trains a transformer-based encoder by reconstructing multisensory observations from only a subset of sensor embeddings, leading to cross-modal prediction and sensor fusion. For downstream policy learning, we introduce a novel asymmetric architecture, where a cross-attention mechanism allows the critic to extract dynamic, task-specific features from the frozen embeddings, while the actor receives a stable pooled representation to guide its actions. Our method demonstrates accelerated learning and robust performance under diverse perturbations, including sensor noise, and changes in object dynamics. Evaluations in multiple challenging, contact-rich robot manipulation tasks in simulation and the real world showcase the effectiveness of MSDP. Our approach exhibits strong robustness to perturbations and achieves high success rates on the real robot with as few as 6,000 online interactions, offering a simple yet powerful solution for complex multisensory robotic control. Website: https://msdp-pearl.github.io/
arXiv:2511.15592v2 Announce Type: replace Abstract: It is well-known that general bilevel linear programs (BLPs) are strongly $NP$-hard, even when the leader's and the follower's objective functions are exact opposites. However, the complexity classification of BLPs remains incomplete when one of the decision-makers has a fixed number of variables or constraints. In this paper, we close the remaining gap in this complexity landscape. Thus, while optimistic BLPs are known to be polynomially solvable when the number of follower variables is fixed, we prove that the corresponding pessimistic problem is strongly $NP$-hard. To the best of our knowledge, this is the first result demonstrating that, under comparable assumptions, the pessimistic formulation can be computationally harder than its optimistic counterpart. In addition, we prove that BLPs remain polynomially solvable in both the optimistic and the pessimistic settings when the number of follower constraints is fixed. We further investigate whether these polynomial-time solvability results persist for bilevel convex quadratic programs. While the optimistic formulation remains polynomially solvable when the number of follower variables is fixed, we prove that the pessimistic formulation with a fixed number of follower constraints becomes $NP$-hard. In other words, unless $P = NP$, there is a strict complexity gap between bilevel programs with linear and convex quadratic objective functions. Finally, we show that replacing a convex quadratic follower objective with a nonconvex quadratic one renders the optimistic problem $NP$-hard, even when both follower dimensions are fixed.
From Fork-Join to Asynchronous Tasks: Parallelizing Tiled Cholesky Decomposition with OpenMP and HPX
arXiv:2606.11937v1 Announce Type: new Abstract: Fork-join parallelism, popularized by OpenMP, remains the dominant model for shared-memory parallel programming, but its implicit synchronization barriers can penalize algorithms with inhomogeneous workloads. Asynchronous many-task (AMT) runtimes sidestep these barriers by expressing work as a dependency graph of fine-grained tasks. Yet, the actual performance benefit over a carefully written fork-join baseline is rarely quantified. In this work, we introduce Cholesky-Bench and use it to revisit the tiled Cholesky decomposition, a canonical irregular kernel, comparing four parallelization variants of the right-looking algorithm across two runtimes: the OpenMP implementations shipped with GCC and LLVM, and the HPX AMT runtime. The variants span classical fork-join, a collapsed fork-join that exposes additional inner-loop parallelism, synchronous tasking, and asynchronous tasking with explicit data dependencies. We benchmark all eight combinations on a dual-socket 128-core AMD Zen 2 node across multiple tile sizes and problem sizes. Our results show that across all variants, HPX outperforms OpenMP at the optimal tile size by 15%-30%. Specifically, asynchronous HPX tasks are up to 26% faster than their OpenMP counterparts, and exhibit roughly 3.8x smaller task overhead. Furthermore, the collapsed fork-join variants close most of the gap to synchronous tasking. Removing redundant synchronization barriers yields an additional improvement of 7% (OpenMP) to 14% (HPX). A GCC-versus-LLVM comparison further reveals compiler-specific differences in fork-join scheduling and task-creation overheads.
arXiv:2606.11609v1 Announce Type: new Abstract: Stance detection requires identifying an author's position toward a target, often from short-form texts where stance is implicit, indirect, or rhetorically framed. Although large language models (LLMs) achieve strong performance on this task, single-pass prompting can be brittle when multiple interpretations are plausible. Existing aggregation strategies, such as majority voting or self-consistency, improve robustness by combining labels, but they discard the intermediate reasoning needed to resolve conflicting interpretations. We introduce a multi-agent reasoning framework with adaptive worker allocation for stance detection that shifts aggregation from label-level voting to reasoning-level synthesis. The framework employs a Manager-Worker architecture in which a Manager adaptively allocates a variable number of Worker agents based on input complexity. Each Worker analyzes the input from a distinct perspective and produces a reasoning-only explanation without emitting a stance label; the Manager then synthesizes these explanations to produce the final prediction. We evaluate the proposed framework on SemEval-2016, P-Stance, and COVID-19 Stance using Llama, Mistral, and Gemini. Results show that the framework yields the largest gains on implicit and context-dependent stance cases, achieving 86.07 Macro-F1 on COVID-19 and 82.90 on SemEval-2016, while remaining competitive on more explicit stance datasets such as P-Stance. These findings suggest that adaptive reasoning-level aggregation is most beneficial when stance cannot be reliably inferred from surface cues alone.
arXiv:2511.16672v4 Announce Type: replace Abstract: Recent advances in large multimodal models (LMMs) have enabled impressive reasoning and perception abilities, yet most existing training pipelines still depend on human-curated data or externally verified reward models, limiting their autonomy and scalability. In this work, we strive to improve LMM reasoning capabilities in a purely unsupervised fashion (without any annotated data or reward distillation). To this end, we propose a self-evolving framework, named EvoLMM, that instantiates two cooperative agents from a single backbone model: a Proposer, which generates diverse, image-grounded questions, and a Solver, which solves them through internal consistency, where learning proceeds through a continuous self-rewarding process. This dynamic feedback encourages both the generation of informative queries and the refinement of structured reasoning without relying on ground-truth or human judgments. When using the popular Qwen2.5-VL as the base model, our EvoLMM yields consistent gains upto $\sim$3\% on multimodal math-reasoning benchmarks, including ChartQA, MathVista, and MathVision, using only raw training images. We hope our simple yet effective approach will serve as a solid baseline easing future research in self-improving LMMs in a fully-unsupervised fashion. Our code and models are available at https://github.com/mbzuai-oryx/EvoLMM.
arXiv:2511.17546v2 Announce Type: replace Abstract: We present a novel method for accurately measuring the absolute electric quadrupole moments of light transition elements $(23 \leq Z \leq 30 )$. Our approach is based on performing precision muonic x-ray spectroscopy of the $2s-2p$ manifold, which is also referred to as the Lamb shift. These transitions are too weak to be detected with dispersive methods and too overlapping to be resolved by solid-state detectors. Here, we propose the use of cryogenic microcalorimeters, which possess high efficiency and excellent energy resolution in the relevant energy regime, coupled with state-of-the-art theoretical calculations. We demonstrate the feasibility of this approach by performing extensive calculations and realistic simulations. In this way, we establish that the uncertainty in the absolute moment, which is transferred to the quadrupole moments of all isotopes in the chain, could be reduced by up to an order of magnitude within a day of measurement. These precise reference quadrupole moments serve as valuable inputs for nuclear structure studies and for benchmarking state-of-the-art quantum chemistry calculations in open-shell elements.
arXiv:2511.18235v3 Announce Type: replace Abstract: The rapid growth of the Internet of Things (IoT) has created large-scale, heterogeneous ecosystems that are increasingly vulnerable to sophisticated, distributed cyber threats. However, many existing anomaly detection systems prioritize detection accuracy while overlooking system-level constraints, such as latency, computational overhead, and energy consumption, thereby limiting their practicality for resource-constrained edge gateways. This paper presents EcoDefender, an edge-oriented hybrid anomaly detection framework that combines Autoencoder (AE)-based latent representation learning with Isolation Forest (IF) anomaly scoring for IoT traffic analysis. The proposed architecture introduces several enhancements over conventional AE-IF pipelines, including anomaly-aware latent manifold regularization, variance-weighted isolation splits in the latent space, and a learnable fusion mechanism that adaptively combines reconstruction error and isolation-based anomaly scores in the presence of potential distributional drift. By compressing high-dimensional traffic features into compact latent representations and performing anomaly scoring in this reduced space, EcoDefender enables lightweight and fully unsupervised anomaly detection suitable for edge deployment. An experimental evaluation of realistic IoT traffic and a distributed Raspberry Pi edge testbed demonstrates that EcoDefender achieves up to 94% detection accuracy while maintaining low computational overhead, with an average CPU usage of 22% and an end-to-end inference latency of 27 ms. Furthermore, energy-aware measurements obtained through device-level power monitoring show an average energy consumption of 0.45 J per inference (0.28 g CO2 emissions), representing a 30% reduction in energy consumption compared with AE-only baselines while sustaining inference throughput of up to 5,000 samples per second.
arXiv:2512.08211v2 Announce Type: replace Abstract: Large language models (LLMs) are moving from cloud-centric services toward on-device embedded AI, where models interact with private, longitudinal signals sensed from users and their physical environments. Mobile phones are a natural platform for such applications because they are continuously carried by users, connected to wearable sensors, and deeply integrated with daily mobile applications. However, practical LLM fine-tuning on commodity phones remains difficult. Existing fine-tuning frameworks are largely Python-based and server-oriented, making them hard to deploy inside mobile applications. We present MobileFineTuner, a mobile-native open-source framework for end-to-end LLM fine-tuning on commodity mobile phones. MobileFineTuner is implemented in C++ and provides a reusable training stack. To make fine-tuning feasible under mobile resource constraints, MobileFineTuner integrates a resource-aware training runtime with memory-efficient attention, activation checkpointing, gradient accumulation, parameter sharding, and energy-aware scheduling. We evaluate MobileFineTuner on real mobile phones using GPT-2, Gemma 3, and Qwen2.5 models across multiple fine-tuning tasks. The results show that MobileFineTuner reproduces standard Full-FT and LoRA fine-tuning behavior, substantially reduces memory pressure and improves executability on memory-constrained phones. We further demonstrate MobileFineTuner through a private campus health-agent application, where a local LLM is fine-tuned on user-specific wearable-sensing records to provide more personalized responses while keeping raw records on the phone. These results establish MobileFineTuner as a practical toolkit for studying and building on-device LLM fine-tuning applications in embedded AI and sensing systems.
arXiv:2512.11393v2 Announce Type: replace Abstract: Humans can intuitively parallelise complex activities, but can a model predict this from observing a single person? Given one egocentric video, we introduce the N-Body Problem: predicting how N individuals, can hypothetically perform the same set of tasks. The goal is to maximise speed-up, but naive assignment of video segments to individuals often violates real-world constraints, leading to physically impossible scenarios like two people using the same object or occupying the same space. To quantify this, we formalise the N-Body Problem and propose a suite of metrics to evaluate both performance (speed-up, task coverage) and feasibility (spatial collisions, object conflicts and causal constraints). As a proof of concept, we introduce a structured prompting strategy that guides a Vision-Language Model (VLM) to reason about the 3D environment, object usage, and temporal dependencies, producing a viable parallel execution. On 100 videos from EPIC-Kitchens and HD-EPIC, for $N = 2$, our structured prompt improves action coverage by 45% over a baseline prompt for Gemini 2.5 Pro, while simultaneously slashing collision rates, object and causal conflicts by 51%, 52% and 55% respectively.
arXiv:2512.12580v2 Announce Type: replace Abstract: This article introduces a novel cryptographic paradigm based on nonderived polyadic algebraic structures. Traditional cryptosystems rely on binary operations within groups, rings, or fields, whose well-understood properties can be exploited in cryptanalysis. To overcome these vulnerabilities, we propose a shift to polyadic rings, which generalize classical rings by allowing operations of higher arity: an $m$-ary addition and an $n$-ary multiplication. The foundation of our approach is the construction of polyadic integers -- congruence classes of ordinary integers endowed with such $m$-ary and $n$-ary operations. A key innovation is the parameter-to-arity mapping $\Phi(a,b)=(m,n)$, which links the parameters $(a,b)$ defining a congruence class to the specific arities required for algebraic closure. This mapping is mathematically intricate: it is non-injective, non-surjective, and multivalued. This complex, non-unique relationship forms the core of the proposed cryptosystem's security. We present two concrete encryption procedures that leverage this structure by encoding plaintext within the parameters of polyadic rings and transmitting information via polyadically quantized analog signals. In one method, plaintext is linked to the additive arity $m_{i}$ and secured using the summation of such signals; in the other, it is linked to a ring parameter $a_{i}$ and secured using their multiplication. In both cases, the "quantized" nature of polyadic operations generates systems of equations that are straightforward for a legitimate recipient with the correct key but exceptionally difficult for an attacker without it. The resulting framework promises a substantial increase in cryptographic security. This work establishes the theoretical foundation for this new class of encryption schemes and highlights their potential for constructing robust, next-generation cryptographic protocols.
arXiv:2606.11662v1 Announce Type: new Abstract: Deep search requires agents to answer complex questions through multi-step web search, browsing, evidence comparison, and synthesis. A central challenge is deciding how to search when several directions look plausible but only some will later lead to reliable evidence. If an agent greedily follows the current best-looking direction, it may keep extending a weak continuation. If it explores without discipline, it may waste budget on disconnected trials. We propose TreeSeeker, an inference-time framework for controlled trial-and-error in deep search. TreeSeeker organizes search as branch-and-return search over tree-structured states, where each branch is a tentative direction for a sub-goal. At each round, TreeSearch reads all sub-goal trees, identifies active goals, and uses textual UCB signals of value, uncertainty, and risk to select among exploiting a promising branch, exploring an uncertain alternative, or pruning an unproductive continuation and returning to an earlier branch point. TreeMem supports this control loop by keeping evidence, uncertainty, conflicts, progress, and failure cues attached to the branches that produced them, so trial outcomes can guide later decisions. Experiments on XBench-DeepSearch, BrowseComp, and BrowseComp-ZH show that TreeSeeker consistently outperforms strong open-source baselines, suggesting that explicit branch-and-return control complements stronger reasoning and tool execution.
arXiv:2606.11663v1 Announce Type: new Abstract: Accurate salary prediction is critical for bridging the information gap between employers and job seekers in modern labor markets. Existing approaches predominantly yield a single point estimate and treat job attributes such as location, occupation, and industry as independent categorical features, ignoring both the inherent uncertainty and multi-modality of real-world compensation data and the rich hierarchical and semantic-similarity relationships that govern pay norms. In this paper we propose GAT-MDN, a unified framework that addresses both limitations simultaneously. For each of the three attribute domains we construct a domain-specific graph whose edges encode (i) hierarchical parent-child containment and (ii) weighted similarity links derived from a pre-trained Sentence-Transformer. Parallel Graph Attention Networks (GATs) with edge-feature-aware attention learn rich, context-sensitive node representations from these multi-relational graphs. A priority-based hierarchical selection module then assembles a composite feature vector that gracefully handles missing or coarse attributes, and a Mixture Density Network (MDN) head maps this vector to the parameters of a Gaussian Mixture Model (GMM), yielding a full conditional salary distribution. Extensive experiments on a real-world Dutch job-posting dataset of over 1 million records demonstrate that GAT-MDN significantly outperforms a non-graph MLP-MDN baseline in both Negative Log-Likelihood (NLL) and Mean Squared Error (MSE).
arXiv:2601.04710v2 Announce Type: replace Abstract: Fine-tuning large language models (LLMs) achieves strong performance but is often limited by the memory overhead of backpropagation. Zeroth-order (ZO) optimization avoids this overhead by estimating gradients through forward passes alone, yet it typically converges slowly because random Gaussian perturbations yield high-variance gradient estimates in high-dimensional parameter spaces. In this paper, we propose a plug-and-play framework that turns random perturbations into more effective descent directions. The key idea is to draw a small pool of candidate perturbations, evaluate their loss values, and then select or combine those that are best aligned with the optimization objective. We develop two instantiations of this idea: MeZO-GV, which forms a guiding vector from the contrast between low-loss and high-loss perturbation groups, and MeZO-Greedy, which keeps the single best perturbation within a fixed evaluation budget. We theoretically show that both strategies yield a larger per-step reduction in the objective than standard ZO estimation, leading to improved convergence rates. Experiments on LLMs of different scales and architectures confirm that the proposed methods integrate naturally with existing ZO optimizers and consistently improve convergence speed and task accuracy. On OPT-13B, our approach outperforms all ZO baselines across 11 benchmarks and exceeds gradient-based methods on 9 of them, while retaining the memory efficiency of forward-only optimization.
arXiv:2606.11667v1 Announce Type: new Abstract: Sybil attacks create an illusion of traffic congestion by utilizing fake identities, which undermines the reliable and safe operation of vehicular ad hoc networks (VANETs). Existing detection mechanisms struggle to effectively handle Sybil attacks as they are (i) susceptible to high false positive rates (FPR) due to the overlapping trajectories of both Sybil and legitimate vehicles, (ii) not practical for real-world deployment due to manual calibrations with ground data, (iii) ineffective for sparse distribution of roadside units (RSUs) and vehicles as they depend heavily on the presence of both, and (iv) inefficient due to computational overheads. This paper addresses these shortcomings and proposes a robust framework to tackle these issues. The proposed scheme reduces the FPR by utilizing GPS location data, enabling the construction of more accurate and distinguishable trajectories. Besides, it employs DBSCAN clustering to identify Sybil vehicles, facilitating unsupervised parameter selection. GPS data eliminates the dependency on RSUs and vehicles, making this scheme effective in both sparse and dense regions. Additionally, the proposed scheme is lightweight and consistent across vehicles with heterogeneous capacities. Experimental results demonstrate that the proposed scheme reduces the FPR by approximately 68% in dense regions and 70% in sparse areas. Furthermore, it lowers the false negative rate (FNR) by 67% in the sparse region and achieves a competitive detection rate compared to the existing methods in both dense and sparse regions. Additionally, the proposed scheme decreases the detection time by almost 80% in dense regions and 43% in sparse ones.
arXiv:2606.11279v1 Announce Type: cross Abstract: Automatic speech recognition systems have been shown to under-perform when it comes to transcribing words rarely seen in the training data, namely specialized terminology. Open-vocabulary keyword spotting, combined with contextual biasing, has been shown to mitigate this issue. However, existing systems can only handle glossaries of a few hundred terms without becoming an infeasible bottleneck. We propose a system that stores features with a memory footprint up to 128 times smaller than a comparable baseline and allows users to process massive databases while remaining open-vocabulary. Without fine-tuning the speech recognition model, our system achieves a comparable entity recall as uncompressed solutions, even in languages not seen during training.
arXiv:2606.11669v1 Announce Type: new Abstract: Generative AI (GenAI) tools offer increasing opportunities for augmenting human cognitive tasks. Among these tasks, information seeking is being rapidly reshaped by GenAI tools, with potentially profound implications for learning and knowledge acquisition. To investigate these implications, we conducted a between-subjects field experiment in which participants pursued informal learning by seeking information through either ChatGPT or Google Search over a span of 8 days. Using a daily diary protocol, we gathered in-situ data on their information-seeking processes. Our findings show that participants in the ChatGPT group experienced diminished agency in their information-seeking processes, as they offloaded much of the information selection to AI, and consequently experienced greater meta-cognitive load arising from this reduced sense of control. We further highlight two sources of distortion in information access when using ChatGPT: biases in ChatGPT outputs, particularly towards providing solution-oriented artifacts over principled knowledge; and systematic shifts in users' information-seeking behaviors, whereby the conversational and socially-oriented interaction paradigm of current GenAI tools may inadvertently reduce exploration of the broader knowledge space. As a result, on average, participants in the ChatGPT group had worse learning outcomes than those using Google, especially for higher-order critical learning. Our work suggests inherent tensions between offloading information seeking to AI and meaningful learning, and provides broader implications for understanding AI's risks to human cognition.
arXiv:2606.11671v1 Announce Type: new Abstract: Agent skills let LLM agents reuse instructions, resources, tools, and workflows, but they also create a new place for malicious behavior to hide. A skill may look benign in its documentation or code while becoming harmful only when it is invoked with particular user requests, local assets, persistent state, or multi-step tool interactions. This makes purely static vetting brittle. We present Runtime Skill Audit (RSA), a dynamic analysis method that audits skills by asking what the skill-mediated agent actually does under targeted runtime conditions. Instead of testing every skill with the same generic tasks, RSA profiles risk-relevant interfaces, prepares the execution context needed to exercise them, and assigns security labels from the resulting trace evidence. We instantiate RSA on OpenClaw and evaluate it on 100 skills against representative static baselines. RSA achieves 90.0\% accuracy with an 88.0\% true positive rate and an 8.0\% false positive rate, improving accuracy by 13.0 percentage points over the best static baseline. Under self-evolving attacks, static detectors collapse after one or two rounds, while RSA continues to detect 19--20 out of 20 malicious skills across rounds.
arXiv:2601.06898v2 Announce Type: replace Abstract: We introduce a stressor-agnostic Resilience Key Performance Indicator (Resilience KPI) for megawatt charging stations (MSC) serving heavy-duty vehicles. Beyond routine performance statistics (e.g., availability, throughput), the KPI quantifies a site's ability to anticipate, operate under degradation, and recover from disruptions using observable signals already in the framework: ride-through capability, restoration speed, service under N-1, expected unserved charging energy, and queue impacts. The headline score is normalised to 0-100 for fair cross-site and cross-vendor benchmarking, with optional stressor-specific breakouts (grid, ICT, thermal, flooding, on-site incidents) for diagnostics and robustness checks. DATEX II provides a solid baseline for resilience KPIs centred on infrastructure inventory, status, and pricing, while additional KPIs, especially around grid capacity, on-site flexibility, heavy-vehicle geometry, environmental hardening, maintenance, and market exposure, are essential for a complete resilience picture and will require extensions or complementary data sources. The KPI is designed for monthly/quarterly reporting to support design and operational decisions and cost-benefit assessment of mitigations (e.g., backup power, spares, procedures). It offers a consistent, transparent methodology that consolidates heterogeneous logs and KPIs into a single, auditable indicator, making resilience comparable across sites, vendors, and jurisdictions.
arXiv:2606.11675v1 Announce Type: new Abstract: Diagnosing pulmonary diseases requires integrating heterogeneous evidence amid phenotypic variability and cross-disease overlap. Although large language models (LLMs) have shown progress on pulmonary knowledge question answering (QA) and information-processing tasks, reliable pulmonary diagnosis requires patient-specific, relation-aware reasoning over electronic medical record (EMR) evidence rather than isolated knowledge recall. We define this gap between pulmonary knowledge and case-level diagnostic reasoning as the Pulmonary Knowledge-to-Diagnosis Gap. To address it, we introduce LungKG, the first structured pulmonary knowledge graph for diagnostic knowledge organization and record-grounded reasoning. LungKG contains 59,038 nodes and 164,308 edges across 15 entity types and 112 relation types, serving as both a reusable pulmonary knowledge resource and the foundation for LungKG-guided model adaptation. Built on LungKG, we propose Lung-R1, a LungKG-guided pulmonary LLM trained through KG-constrained reasoning-chain construction and KG-guided reinforcement learning. In a 20-system evaluation, Lung-R1-14B achieves state-of-the-art performance across Choice, Pulmonary-QA, and EMR Diagnosis, reaching an EMR Diagnosis score of 4.3583 and surpassing the strongest non-Lung-R1 baseline by 0.1476 points. These results demonstrate the value of LungKG-guided training for EMR-based pulmonary diagnosis.
arXiv:2606.11977v1 Announce Type: new Abstract: In this report, we present our third-place solution for the DataMFM Challenge Track 1: Document Parsing. This track requires models to recover structured Markdown documents from document page images while preserving textual content and document structure. To address the complementary requirements of accurate content recovery and faithful structure reconstruction, we propose ParseFixer, an agentic framework for backbone parsing and selective correction. ParseFixer consists of two key modules: Full-Page Backbone Parsing (FBP) and Agentic Selective Correction (ASC). FBP produces stable initial Markdown outputs with MinerU2.5 Pro, while ASC detects high-value parsing failures and repairs them through a verify-and-rollback correction process. By placing selective multimodal correction after open-source backbone parsing, ParseFixer improves the recovery of key document elements without rewriting reliable backbone predictions. On the test set, our final system achieves an overall score of 61.78 and ranks third in Track 1, demonstrating its effectiveness for accurate document parsing. Our code will be released at: https://github.com/iLearn-Lab/CVPRW26-ParseFixer.
arXiv:2601.10724v4 Announce Type: replace Abstract: Multi-robot formation control has various applications in domains such as vehicle troops, platoons, payload transportation, and surveillance. Maintaining formation in a vehicle platoon requires designing a suitable control scheme that can tackle external disturbances and uncertain system parameters while maintaining a predefined safe distance between the robots. A crucial challenge in this context is dealing with the unknown/uncertain friction forces between wheels and the ground, which vary with changes in road surface, wear in tires, and speed of the vehicle. Although state-of-the-art adaptive controllers can handle a priori bounded uncertainties, they struggle with accurately modeling and identifying frictional forces, which are often state-dependent and cannot be a priori bounded. This thesis proposes a new adaptive sliding mode controller for wheeled mobile robot-based vehicle platoons that can handle the unknown and complex behavior of frictional forces without prior knowledge of their parameters and structures. The controller uses the adaptive sliding mode control techniques to regulate the platoon's speed and maintain a predefined inter-robot distance, even in the presence of external disturbances and uncertain system parameters. This approach involves a two-stage process: first, the kinematic controller calculates the desired velocities based on the desired trajectory; and second, the dynamics model generates the commands to achieve the desired motion. By separating the kinematics and dynamics of the robot, this approach can simplify the control problem and allow for more efficient and robust control of the wheeled mobile robot.
arXiv:2606.11676v1 Announce Type: new Abstract: Traditional wildfire models rely on rigid, low-dimensional parameters and static fuel maps, frequently underpredicting fire spread. To address this weakness, we introduce a hybrid deep-learning parameterized Probabilistic Cellular Automata (CA) framework implemented in JAX. Our approach employs a Multi-Scale Convolutional Neural Network to dynamically generate spatially varying parameters that govern fire-spread probability, wind alignment, and slope influence. This hybrid design captures complex, nonlinear environmental interactions while preserving the physical interpretability of the underlying three-state CA. The JAX implementation enables hardware acceleration and gradient-based parameter calibration. Evaluated on six large-scale wildfires in the western United States, the model maintains IoU > 0.6 over 72-hour forecast horizons after a 10-day data assimilation window during which the model is fitted incrementally to observed perimeters; the resulting forecast is a conditional projection of fire growth under the suppression regime already ncoded in those observations.
arXiv:2606.11680v1 Announce Type: new Abstract: Large language model (LLM) agents struggle with long-horizon tasks due to their inherent statelessness, requiring all task-relevant information to be encoded in growing input contexts. The resulting degraded reasoning quality, increased inference cost, and higher latency necessitate efficient working memory mechanisms. However, existing approaches either rely on lossy compression or similarity-based retrieval, which often fail to capture temporal structure and causal dependencies required for multi-step agentic tasks. In this work, we present HORMA, a Hierarchical Organize-and-Retrieve Memory Agent that organizes experience into a file-system-like hierarchical structure, where summarized entities are linked to the corresponding raw trajectories, enabling efficient access without losing detailed information. HORMA decomposes working memory into two stages: structured memory construction and navigation-based retrieval. The construction module iteratively refines how experiences are structured by distinguishing between failures caused by missing information and those caused by misleading or overloaded context. The navigation module retrieves task-relevant context by traversing the hierarchy using a lightweight agent trained with reinforcement learning to select minimal yet sufficient context, thereby reducing latency along the critical execution path. Across ALFWorld, LoCoMo, and LongMemEval, HORMA improves task performance under constrained context budgets while requiring at most 22.17% of the baseline token usage in long conversation tasks. Compared to existing methods, it consistently achieves better efficiency-performance trade-offs and generalizes effectively to unseen tasks.