Search arXiv⌕ Search

arXiv subjects

Can Xu

Publications and source records attributed to Can Xu.

At least 19 recordsLinked to original sources
arXiv2026

From Transient Prompts to Persistent Control: Scientific Poster Generation via Recursive Semantic-Geometric Contracts

Scientific poster generation distills a multimodal paper into a single-page visual artifact, forcing strict trade-offs between informational coverage and readability under a fixed spatial budget. Existing methods pass plans as transient prompts and validate individual stages in isolation. This strategy causes requirements to drift across content and layout modules, and previous checks to be silently invalidated. We introduce PosterVisor, a control framework that shifts poster generation from transient prompts to persistent control. An Orchestrator grounds rubrics in the paper and visual assets, compiling them into a Semantic-Geometric Contract (SGC) that binds claims and sources to required visuals, budgets, and spatial commitments. Only fully instantiated records become executable assertions; other usable requirements remain soft guidance. Recursive Contract Enforcement (RCE) dynamically triggers checks across stages as evidence emerges. Crucially, during repairs, RCE rechecks affected checkpoint states, preventing repair-induced regressions from propagating silently. We instantiate PosterVisor in HTML/CSS and editable PPTX generators. On the 100-paper Paper2Poster benchmark, PosterVisor-PPT improves observed mean poster-grounded QA accuracy over PosterGen (64.47% vs. 58.53%) and is preferred by human judges in 72.5% of non-tied pairwise comparisons (95% CI, 61.6-83.4%). A secondary 30-paper study also yields higher VLM Overall and PaperQuiz means. These results support rubric-compiled contracts and stage-conditioned enforcement for controllable poster synthesis.

cs.AI↗
arXiv2026

MAC-I$^2$: Learned Metrics-Aware Covariance for Robust Visual-Inertial Fusion in Initialization and Calibration

Visual-Inertial (VI) fusion is fundamental to accurate and robust state estimation, where camera and IMU measurements are combined according to their respective uncertainties. Existing methods, however, fuse the two modalities with predefined uncertainties, regardless of how reliable each is in the local context, and thus often struggle under challenging environments involving illumination changes, dynamic objects, and textureless regions. In this paper, we present MAC-I$^2$, which achieves robust VI fusion through learned metric-aware covariance for both modalities, so that vision and IMU compete on their own merits rather than relying on predefined uncertainties. Here, metrics-aware means that each predicted covariance faithfully reflects the actual magnitude of the corresponding measurement noise. On the visual side, we propagate learned feature-matching uncertainties into pose covariances for the fusion. On the inertial side, motivated by the observation that integration error accumulates sharply at the early stage and grows slowly afterward, we design a learned IMU model with a learnable initial covariance, and propose a dedicated fine-tuning strategy on a held-out training subset to enable the metrics-aware covariance on unseen sequences. As a showcase, we build a VI initialization and calibration system, since accurate and robust initialization and calibration are the prerequisite for any reliable VI system. Experiments on EuRoC, and VBR show that MAC-I$^2$ substantially outperforms existing methods: it achieves a 99.9% initialization success rate on EuRoC, reducing gravity and velocity errors by about 60% and 42% over the strongest baseline, and maintains 80% success rate on challenging VBR sequences where baseline methods such as VINS-Mono drop below 10%.

cs.RO↗
arXiv2026

DuMateBench: Evaluating Autonomous Agents in Complex Real-World Workflows

Autonomous agents are increasingly adopted to complete complex, multi-tool workflows in real-world settings. However, existing benchmarks typically separate tasks by application or capability and evaluate agents in environments that are cleaner and more stable than those encountered in practice. We introduce DuMateBench, a real-session benchmark reconstructed from anonymized and privacy-screened user sessions collected from a large-scale production agent platform. Each task preserves the relevant pre-solution interaction history, persistent configurations, and workspace state, and is then validated through human verification. The resulting benchmark comprises 200 tasks spanning 8 broad scenarios and 17 fine-grained capability categories, with most tasks requiring multiple capability coordination. We execute these tasks in isolated Docker containers injected with three forms of real-world environmental complexity: Insufficient, Unstable, and Noisy, and assess performance using a hybrid deterministic and LLM-as-Judge evaluation protocol. Experiments across five representative autonomous-agent frameworks paired with four state-of-the-art LLMs reveal substantial gaps in strict task completion. Complementary robustness, efficiency, and diagnostic analyses further show that performance under environmental perturbations is jointly shaped by the capabilities of the LLM and the surrounding agent framework. The code and data are publicly available at https://dumatebench.com/.

cs.AI↗
arXiv2026

Dust in the Average Galaxy: Attenuation, Emission, and Opacity from $0<z<7$

We present constraints on the dust emission and attenuation properties of galaxies across 0 10^{10.5}M_\odot$. We derive empirical relationships for the effective attenuation, dust temperature, fraction of star formation that is unobscured, and dust-to-stellar mass ratio as functions of redshift and stellar mass. We separate the first order effect of star/dust geometry from dust grain properties by combining constraints on the IR SED, UV SED, and dust mass surface density. Importantly, we measure over an order of magnitude decrease in $κ_{UV}/κ_{FIR}$--the ratio of dust mass absorption coefficients in the UV at 1600Å and FIR at 500$μ$m--from z~0 to z~7. A depressed $κ_{UV}/κ_{FIR}$ is consistent with a deficit of small dust grains, possibly attributable to the intense radiation fields of high-$z$ star formation; indeed, we find a redshift-invariant inverse relationship between $κ_{UV}/κ_{FIR}$ and $Σ_{SFR}$. Most evolution in the dust-to-stellar ratio is at $z<1$, the product of mild downward evolution in the dust-to-gas ratio combined with steep evolution in the gas-to-stellar ratio. The significant evolution and dynamic range of $κ_{UV}/κ_{FIR}$ and prevailing disconnect between the UV/optical and FIR regimes emphasize that direct dust constraints are irreplaceable for the majority of star-forming galaxies at z<7, not just the most extreme star-formers.

astro-ph.GA↗
arXiv2026

PNEC-Mamba: Prototype-Guided Positive-Negative Evidence Calibration for Hyperspectral Image Classification

In real-world hyperspectral scenes, pixel representations are often ambiguous due to factors such as spectral similarity, mixed pixels, and local context interference, which may simultaneously encode discriminative evidence and interfering information. Existing methods mainly focus on learning more powerful representations or modeling broader contexts, but rarely investigate whether the learned representations provide reliable evidence or introduce interference into classification decisions. To address this issue, we view hyperspectral image classification from the perspective of pixel-level evidence reliability modeling and propose PNEC-Mamba, a prototype-guided positive-negative evidence calibration framework. The framework progressively establishes semantic references, separates class-related evidence from interference, estimates pixel-level reliability, and performs selective calibration. First, a full-image state-space encoder extracts pixel representations, while dynamic class prototypes provide semantic references that evolve jointly with the feature space. Subsequently, positive and negative evidence is derived from pixel-prototype competition, explicitly separating discriminative cues that support classification from confusing signals associated with competing classes. Based on these evidence relationships, a multi-source uncertainty estimation strategy is introduced to assess pixel-level reliability, enabling stronger evidence calibration for uncertain regions. Finally, a full-resolution consistency refinement step is applied to recover local spatial details and improve boundary coherence in the final predictions. Extensive experiments on three benchmark datasets demonstrate that PNEC-Mamba achieves superior classification performance compared with state-of-the-art methods.

cs.CV↗
arXiv2026

From Starburst to Quenching: Physical Properties of Extremely Compact Starbursts at z$\sim$0.1

The compaction phase plays a crucial role in galaxy evolution, as it is strongly linked to star formation activities and structural transformation. We have identified a sample of extremely compact starburst galaxies (eCSBs) at low redshift~(z$\sim$0.1), which represent this critical evolutionary stage. These eCSBs are massive outliers with intense star formation and high infrared luminosities comparable to (U)LIRGs, while their structure already resembles quiescent galaxies. To investigate their molecular gas properties, we conducted IRAM 30m observations of $^{12}$CO J = 1--0 and $^{12}$CO J = 2--1 emission lines. Our results indicate that eCSBs exhibit a notably low molecular gas fraction~($\sim3\%$), and short gas depletion time~($\sim$ 20 Myr), suggesting that these galaxies are rapidly exhausting their remaining gas reservoir. Compared to normal (U)LIRGs, eCSBs show systematically lower $^{12}$CO(2-1)/$^{12}$CO(1-0) ratio~($R_{21} \sim 0.65 \pm 0.06$), similar to main sequence galaxies. The relatively low CO excitation may be associated with their high central stellar mass densities. These findings provide new insight into the molecular gas properties of galaxies during the compaction phase, highlighting their unique condition and rapid evolution toward quiescence.

astro-ph.GA↗
arXiv2026

TurnOPD: Making On-Policy Distillation Turn-Aware for Efficient Long-Horizon Agent Training

On-policy distillation (OPD) trains a student policy by matching a stronger teacher on the student's own trajectories, offering a promising framework for language agent training. However, its application to long-horizon agentic tasks remains insufficiently explored. We identify two key inefficiencies in vanilla agent OPD: (1) full-horizon rollouts often waste wall-clock resources on tail turns that provide weak and noisy KL supervision, and (2) trajectory-level KL objectives concentrate most of the loss on shallow tokens, leaving deeper decision turns under-trained once initial behaviors are aligned. To address these challenges, we propose TurnOPD, a turn-level budgeting strategy for efficient on-policy distillation of long-horizon agents. TurnOPD consists of two budget controllers: adaptive rollout-depth budgeting, which uses probe-based turn statistics to determine rollout length, and progressive turn-normalized loss budgeting, which gradually shifts KL weighting from token-level to turn-balanced supervision. Experiments on ALFWorld, WebShop, and Multi-Hop Search with task-specialized teacher models show that TurnOPD achieves superior validation accuracy under equal wall-clock training budgets and advances the accuracy--time frontier beyond vanilla OPD.

cs.AI↗
arXiv2026

Playing 20 Question Game with Policy-Based Reinforcement Learning

The 20 Questions (Q20) game is a well known game which encourages deductive reasoning and creativity. In the game, the answerer first thinks of an object such as a famous person or a kind of animal. Then the questioner tries to guess the object by asking 20 questions. In a Q20 game system, the user is considered as the answerer while the system itself acts as the questioner which requires a good strategy of question selection to figure out the correct object and win the game. However, the optimal policy of question selection is hard to be derived due to the complexity and volatility of the game environment. In this paper, we propose a novel policy-based Reinforcement Learning (RL) method, which enables the questioner agent to learn the optimal policy of question selection through continuous interactions with users. To facilitate training, we also propose to use a reward network to estimate the more informative reward. Compared to previous methods, our RL method is robust to noisy answers and does not rely on the Knowledge Base of objects. Experimental results show that our RL method clearly outperforms an entropy-based engineering system and has competitive performance in a noisy-free simulation environment.

cs.HC↗
arXiv2026

VeriEvol: Scaling Multimodal Mathematical Reasoning via Verifiable Evol-Instruct

Scaling reinforcement learning for visual mathematical reasoning requires more than generating harder questions: as data volume grows, the reward labels themselves must remain reliable. Yet existing data pipelines scale supervision while trusting the labeller, and policy-side methods assume the underlying answers are already correct. We instead treat scaling as a verifiable data-construction problem and decouple two axes before any policy update: prompt difficulty, expanded by route-specific evolution operators, and answer reliability, enforced by offline hypothesis-test falsification. We instantiate this as VeriEvol, an iterative framework with two extensible components: a type-aware evolution module that rewrites low-difficulty image-question seeds into harder, image-grounded prompts; and HTV-Agent, a verifier that accepts an answer only after multi-source counter-evidence has failed to refute it. The resulting verified data scales in volume, extends by adding evolution routes or verifier channels, and plugs directly into existing GRPO-style RL recipes. On a five-benchmark visual-math suite, scaling evolved SFT data from 10K to 250K samples raises the mean accuracy from 35.42 to 54.73; then, with backbone, SFT initialization, and GRPO recipe held fixed, VeriEvol adds a cumulative +3.88 over an un-evolved RL baseline, of which +1.82 comes from evolved prompts and +2.06 from the HTV-Agent verifier. We release the prompts, data, models, code, and the full verifier trace of every sample, so that downstream work can scale and audit the pipeline rather than only inspect its outputs.

cs.AI↗
arXiv2026

Ram-pressure stripping caught in action in a young cluster at $z = 2.51$

Galaxy clusters in the local Universe are dominated by massive quiescent galaxies with old ages, formed at high redshifts. Whether their quenching is driven by internal processes or environmental effects is a matter of debate that has been challenging to resolve due to the lack of observations during their peak formation epoch. Here we report clear evidence from the Atacama Large Millimeter/submillimeter Array of extended and elongated gas tails in five galaxies in a forming cluster at z = 2.51. The single-tailed gas distributions, which extend notably beyond the stellar emission probed by JWST in galaxies that are relatively isolated and lack signatures of mergers or interactions (features that are very uncommon in the field), provide evidence of ram-pressure stripping. These very distant confirmed cases of ram-pressure stripping highlight the critical role of environmental effects in gas removal at high redshifts, an often-overlooked quenching pathway.

astro-ph.GA↗
arXiv2026

STARE: Surprisal-Guided Token-Level Advantage Reweighting for Policy Entropy Stability

Reinforcement Learning with Verifiable Rewards algorithms like GRPO have emerged as the dominant post-training paradigm for complex reasoning in LLMs, yet commonly suffer from policy entropy collapse during training. We conduct a first-order gradient analysis of token-level entropy dynamics under GRPO and identify a token-level credit assignment mismatch: the per-token entropy variation decomposes into the product of the trajectory-level advantage and an entropy sensitivity function over the next-token distribution, yielding an advantage-surprisal four-quadrant structure and a near-criticality property. Motivated by it, we propose STARE (Surprisal-guided Token-level Advantage Reweighting for policy Entropy stability), which identifies entropy-critical token subsets via batch-internal surprisal quantiles, selectively reweights their effective advantages, and incorporates a target-entropy closed-loop gate for stable entropy regulation. Across model scales from 1.5B to 32B and three task families (Short CoT, Long CoT, and Multi-Turn Tool Use), STARE sustains stable RL training over thousands of steps while maintaining policy entropy within the target band. On AIME24 and AIME25, STARE outperforms DAPO and other competitive baselines by 4%-8% in average accuracy, with reflection tokens and response length growing in tandem, indicating sustained exploration-exploitation balance that further unlocks RL training potential.Code is available at https://github.com/hp-luo/STARE.

cs.LG↗
arXiv2026

DuMate-DeepResearch: An Auditable Multi-Agent System with Recursive Search and Rubric-Grounded Reasoning

Deep Research (DR) has emerged as a new agentic paradigm to tackle complex, open-ended research tasks, demanding systems that can iteratively frame problems, acquire evidence, verify sources, and synthesize long-form reports. In practice, however, current DR systems are constrained by four interrelated limitations: long-horizon planning over an underspecified scope, the bottleneck of decomposing and scheduling such tasks within a single agent, hallucination risk in long-form synthesis, and limited process auditability. This technical report presents DuMate-DeepResearch, a multi-agent DR framework built on the Qianfan Agent Foundry. The framework decouples the Agent Core, which handles task understanding, planning, and scheduling, from an extensible Tool Ecosystem for retrieval, evidence acquisition, and report rendering, making every intermediate decision and tool invocation explicitly traceable. Building on this infrastructure, DuMate-DeepResearch further introduces three mechanisms: (i) a graph-based dynamic planning strategy expands the research roadmap coarse-to-fine and continuously revises it through reflection, re-planning, backtracking, and parallel branching; (ii) a recursive two-level execution design delegates each complex search sub-task to an inner Search Agent that runs its own planning loop, isolating noisy retrieval and stabilizing long-horizon execution; (iii) a rubric-based test-time optimization mechanism dynamically generates task-specific quality criteria and uses them as live reasoning scaffolds for evidence-grounded synthesis and adaptive stopping. Across two deep research benchmarks, DuMate-DeepResearch establishes new state-of-the-art results: the best overall score (58.03%) on DeepResearch Bench, and the best overall score (61.95%) on DeepResearch Bench II while ranking first in information recall and analysis.

cs.AI↗
arXiv2026

The impact of cosmic filaments on starburst galaxies across cosmic times

Cosmological simulations suggest that various galaxy properties depend on their location within the cosmic web. Yet direct observational evidence of the dependence of star formation activity on distance to filaments remains scarce and is missing at z>1. We investigate how starburst, main-sequence (MS), and quenched galaxies are distributed with respect to cosmic web filaments, and how this distribution evolves with redshift. We first use the SIMBA cosmological simulation to predict the redshift evolution of the mean distance to the closest filament from z=3 to z=0 for different galaxy populations after removing stellar-mass dependencies. We then measure the corresponding signal in the COSMOS field, using COSMOS2020 and COSMOS-Web data, where accurate photometric redshifts enable reconstruction of the projected cosmic web from z=2 to z=0.5, and starbursts are identified through far-infrared spectral energy distribution fitting. In agreement with the results from SIMBA, starburst galaxies are found closer to filaments at z>1 and at larger distances at z<1, MS galaxies occupy intermediate environments with little evolution, and quenched galaxies show progressively shorter distances to filaments toward low redshift, with a crossing between starburst and MS populations around z~1. In COSMOS-Web, the relative evolution in the average distance to filaments between starburst and MS galaxies is detected at a significance level of at least 5σ. We show that a minimal toy model in which the only environmental ingredient is the sSFR-filament distance modulation measured in simulations is sufficient to reproduce the observed differential evolution of the average filament distance between starburst and MS galaxies. These results show that the imprint of large-scale environmental effects on the star formation activity of galaxies, predicted by simulations, is detectable from z=2 down to z=0.5.

astro-ph.CO↗
arXiv2026

Negative Advantages Is a Double-Edged Sword: Calibrating advantages in GRPO for Search Agents

Search agents achieve strong question-answering performance through multi-turn interactions with search engines, with Group Relative Policy Optimization (GRPO) being a widely used training algorithm. However, GRPO-style algorithms still face several challenges in multi-hop search settings. First, correct intermediate steps are often penalized when the final answer is wrong. Second, training is highly unstable, often causing degradation of natural language ability or even catastrophic training collapse. Our analysis attributes these issues to coarse-grained advantage assignment and an imbalance between positive and negative advantages. To address these problems, we propose CalibAdv, an advantage calibration method specifically designed for search agents that enables more accurate and more stable modeling of penalties and rewards. Specifically, CalibAdv leverages the correctness of intermediate steps to downscale excessive negative advantages at a fine-grained level. It then further rebalances positive and negative advantages to improve training stability. Importantly, CalibAdv adopts a lightweight design that calibrates advantages from standard rollout signals, making it simple and easy to deploy. Extensive experiments across three models and seven benchmarks demonstrate that CalibAdv improves both model performance and training stability. Our code is available at https://github.com/wujwyi/CalibAdv.

cs.CL↗
arXiv2026

Do Phone-Use Agents Respect Your Privacy?

We study whether phone-use agents respect privacy while completing benign mobile tasks. This question has remained hard to answer because privacy-compliant behavior is not operationalized for phone-use agents, and ordinary apps do not reveal exactly what data agents type into which form entries during execution. To make this question measurable, we introduce MyPhoneBench, a verifiable evaluation framework for privacy behavior in mobile agents. We operationalize privacy-respecting phone use as permissioned access, minimal disclosure, and user-controlled memory through a minimal privacy contract, iMy, and pair it with instrumented mock apps plus rule-based auditing that make unnecessary permission requests, deceptive re-disclosure, and unnecessary form filling observable and reproducible. Across five frontier models on 10 mobile apps and 300 tasks, we find that task success, privacy-compliant task completion, and later-session use of saved preferences are distinct capabilities, and no single model dominates all three. Evaluating success and privacy jointly reshuffles the model ordering relative to either metric alone. The most persistent failure mode across models is simple data minimization: agents still fill optional personal entries that the task does not require. These results show that privacy failures arise from over-helpful execution of benign tasks, and that success-only evaluation overestimates the deployment readiness of current phone-use agents. All code, mock apps, and agent trajectories are publicly available at~ https://github.com/FreedomIntelligence/MyPhoneBench.

cs.CR↗
arXiv2026

RubricBench: Aligning Model-Generated Rubrics with Human Standards

As Large Language Model (LLM) alignment evolves from simple completions to complex, highly sophisticated generation, Reward Models are increasingly shifting toward rubric-guided evaluation to mitigate surface-level biases. However, the community lacks a unified benchmark to assess this evaluation paradigm, as existing benchmarks lack both the discriminative complexity and the ground-truth rubric annotations required for rigorous analysis. To bridge this gap, we introduce RubricBench, a curated benchmark with 1,147 pairwise comparisons specifically designed to assess the reliability of rubric-based evaluation. Our construction employs a multi-dimensional filtration pipeline to target hard samples featuring nuanced input complexity and misleading surface bias, augmenting each with expert-annotated, atomic rubrics derived strictly from instructions. Comprehensive experiments reveal a substantial capability gap between human-annotated and model-generated rubrics, indicating that even state-of-the-art models struggle to autonomously specify valid evaluation criteria, lagging considerably behind human-guided performance.

cs.AI↗
arXiv2026

AgentMath: Empowering Mathematical Reasoning for Large Language Models via Tool-Augmented Agent

Large Reasoning Models (LRMs) like o3 and DeepSeek-R1 have achieved remarkable progress in reasoning tasks with long cot. However, they remain computationally inefficient and struggle with accuracy when solving problems requiring complex mathematical operations. In this work, we present AgentMath, an agent framework that seamlessly integrates language models' reasoning capabilities with code interpreters' computational precision to efficiently tackle complex mathematical problems. Our approach introduces three key innovations: (1) An automated method that converts natural language chain-of-thought into structured tool-augmented trajectories, generating high-quality supervised fine-tuning (SFT) data to alleviate data scarcity; (2) A novel agentic reinforcement learning (RL) paradigm that dynamically interleaves natural language generation with real-time code execution. This enables models to autonomously learn optimal tool-use strategies through multi-round interactive feedback, while fostering emergent capabilities in code refinement and error correction; (3) An efficient training system incorporating innovative techniques, including request-level asynchronous rollout scheduling, agentic partial rollout, and prefix-aware weighted load balancing, achieving 4-5x speedup and making efficient RL training feasible on ultra-long sequences with scenarios with massive tool invocation. The evaluations show that AgentMath achieves state-of-the-art performance on challenging mathematical competition benchmarks including AIME24, AIME25, and HMMT25. Specifically, AgentMath-30B-A3B attains 90.6%, 86.4%, and 73.8% accuracy respectively, surpassing OpenAI-o3-mini and Claude-Opus-4.0-Thinking while remaining competitive with OpenAI-o3, Gemini-2.5-Pro, and DeepSeek-R1-671B-0528.These results validate the effectiveness of our approach and pave the way for building scalable mathematical reasoning agents.

cs.AI↗
arXiv2026

Beyond Length Scaling: Synergizing Breadth and Depth for Generative Reward Models

Recent advancements in Generative Reward Models (GRMs) have demonstrated that scaling the length of Chain-of-Thought (CoT) reasoning considerably enhances the reliability of evaluation. However, current works predominantly rely on unstructured length scaling, ignoring the divergent efficacy of different reasoning mechanisms: Breadth-CoT (B-CoT, i.e., multi-dimensional principle coverage) and Depth-CoT (D-CoT, i.e., substantive judgment soundness). To address this, we introduce Mix-GRM, a framework that reconfigures raw rationales into structured B-CoT and D-CoT through a modular synthesis pipeline, subsequently employing Supervised Fine-Tuning (SFT) and Reinforcement Learning with Verifiable Rewards (RLVR) to internalize and optimize these mechanisms. Comprehensive experiments demonstrate that Mix-GRM establishes a new state-of-the-art across five benchmarks, surpassing leading open-source RMs by an average of 8.2\%. Our results reveal a clear divergence in reasoning: B-CoT benefits subjective preference tasks, whereas D-CoT excels in objective correctness tasks. Consequently, misaligning the reasoning mechanism with the task directly degrades performance. Furthermore, we demonstrate that RLVR acts as a switching amplifier, inducing an emergent polarization where the model spontaneously allocates its reasoning style to match task demands. The synthesized data and models are released at \href{https://huggingface.co/collections/DonJoey/mix-grm}{Hugging Face}, and the code is released at \href{https://github.com/Don-Joey/Mix-GRM}{Github}.

cs.AI↗