arXiv subjects

Ziyu Liu

Publications and source records attributed to Ziyu Liu.

At least 37 records · Page 2Linked to original sources

arXiv2026

ICCDesign: An R Package for the Design and Analysis of ICC-Based Reliability Studies with Continuous Responses

The intraclass correlation coefficient (ICC) is among the most widely used statistics in reliability research, playing a central role in medical measurement, psychological assessment, and behavioral science. However, practical application of ICC faces two major obstacles. First, ICC can be organized into multiple forms under the McGraw and Wong (1996) framework -- including six widely reported standard forms and four additional design combinations -- and researchers must select the appropriate form based on their study design, yet existing guidelines are not always operationalized in software interfaces. Second, available R tools are highly fragmented: sample size calculation, ICC estimation with confidence intervals, and reliability evaluation are distributed across separate packages, compelling researchers to switch between tools and increasing the risk of analytical errors. This paper introduces the ICCDesign package, designed specifically to provide an integrated workflow for ICC-based reliability studies with continuous responses, assuming one continuous rating per subject-rater cell. The package integrates four core functionalities: (1) point estimation, ANOVA-based confidence intervals, and implemented hypothesis tests for supported ICC design combinations following the McGraw and Wong (1996) framework, with a built-in four-step decision framework guiding users toward an appropriate ICC form; (2) sample size planning based on Zou's (2012) closed-form formulas, supporting two planning modes and an inverse assurance calculation; (3) automated reliability evaluation based on Koo and Li (2016) criteria, with an uncertainty notification when the confidence interval spans the 0.75 good-reliability threshold; and (4) an interactive Shiny web application covering the main analysis and planning functionalities. ICCDesign is available from GitHub at https://github.com/KlariZhang/ICCDesign.

stat.ME↗

arXiv2026

InsightVQA: High-Dimensional Emotion-Cognitive Visual Question Answering Benchmark

Visual emotion understanding requires models not only to recognize emotional states, but also to why they arise and perform higher-level cognitive reasoning. However, existing benchmarks mainly focus on emotion recognition, offering limited support for grounded understanding and response-oriented analysis. To address this gap, we introduce \textbf{InsightVQA}, a large-scale dataset for hierarchical visual question answering on emotion understanding and cognitive reasoning. Building from 351K images collected from six public sources, we apply a rigorous multi-stage filtering pipeline to curate 138K high-confidence images. Each image is annotated at three hierarchical levels: perception QA for emotion and valence recognition, grounded understanding QA constructed from visual trigger extraction through constraint-guided generation, and cognition QA centered on response intent prediction and sequential insight reasoning. In total, InsightVQA contains 725K QA pairs. We further present \textbf{InsightVQA-Bench}, a high-quality evaluation benchmark comprising 30K samples for fine-grained evaluation. To support evaluation, we introduce \textbf{InsightNet}, an emotion-tuned baseline for MLLMs. Results demonstrate that InsightVQA poses significant challenges for grounded emotion understanding and reasoning.

cs.CV↗

arXiv2026

A note on the strong Feller property via the moment method

This note studies the 1D stochastic heat equation driven by a one-dimensional Brownian motion. We prove that the associated Markov process satisfies the strong Feller property under mild non-degeneracy conditions. The approach combines Malliavin calculus with the moment method from PDE control theory.

math.PR↗

arXiv2026

Follow the wobble: Statistical methods to detect astrometric binary asteroids in Gaia FPR

In a previous article, we obtained the first-ever list of astrometric binary asteroid candidates. Some of these candidates have now been confirmed. In that previous work, however, the details of the statistical methods were not provided. Our first aim is to provide methodological details and performance evaluation of the approach used for detecting binaries. Our second aim is to establish an updated list of binary asteroid candidates from Gaia FPR astrometric residuals exploration, where we account for the statistical properties of FPR data. We account for the astrometric uncertainties from FPR and we refine the statistical model of the data, which we use in MC simulation to evaluate the strength of the individual detections; we set up a trend detection method in the residuals and apply a dedicated period search algorithm; we update the statistical selection process to build the list of candidates; we set up a method for detecting objects in multiple windows of consecutive observation; we refine the method for confidence interval estimation of these parameters and we better constrain the physical parameter selection. We detect 343 binary asteroid candidates corresponding to 410 windows of consecutive observations in the astrometric data. We show that in noise-only control simulations, the typical number of detections is 88% lower than in the FPR data. We also detect 9 known binaries, 25 candidates overlapping with the Pan-STARSS survey and 99 overlapping with our previous binary search in DR3. Finally, we report the detection of 45 objects with trends in residuals suggestive of wide binary systems. Our results and analyses demonstrate that although detecting binary asteroids is a difficult problem due to their low signal level, the proposed method is likely to provide a reliable list of detections, including systems poorly accessible to conventional techniques.

astro-ph.EP↗

arXiv2026

Quantitative exponential mixing for the randomized Chirikov standard map

We investigate the mixing properties of a randomized Chirikov standard map on $\mathbb{T}^2$. While the deterministic dynamics exhibit obstructions to global ergodicity, we establish explicit almost-sure quantitative exponential mixing when kicking strengths are sufficiently large. To achieve this, we formulate a criterion for incompressible random dynamical systems, reducing quantitative exponential mixing to serval verifiable conditions. Additionally, we provide a milder parameter condition to derive qualitative exponential mixing and enhanced dissipation.

math.PR↗

arXiv2026

RAR: Retrieving And Ranking Augmented MLLMs for Visual Recognition

CLIP (Contrastive Language-Image Pre-training) uses contrastive learning from noise image-text pairs to excel at recognizing a wide array of candidates, yet its focus on broad associations hinders the precision in distinguishing subtle differences among fine-grained items. Conversely, Multimodal Large Language Models (MLLMs) excel at classifying fine-grained categories, thanks to their substantial knowledge from pre-training on web-level corpora. However, the performance of MLLMs declines with an increase in category numbers, primarily due to growing complexity and constraints of limited context window size. To synergize the strengths of both approaches and enhance the few-shot/zero-shot recognition abilities for datasets characterized by extensive and fine-grained vocabularies, this paper introduces RAR, a Retrieving And Ranking augmented method for MLLMs. We initially establish a multi-modal retriever based on CLIP to create and store explicit memory for different categories beyond the immediate context window. During inference, RAR retrieves the top-k similar results from the memory and uses MLLMs to rank and make the final predictions. Our proposed approach not only addresses the inherent limitations in fine-grained recognition but also preserves the model's comprehensive knowledge base, significantly boosting accuracy across a range of vision-language recognition tasks. Notably, our approach demonstrates a significant improvement in performance on 5 fine-grained visual recognition benchmarks, 11 few-shot image recognition datasets, and the 2 object detection datasets under the zero-shot recognition setting.

cs.CV↗

arXiv2026

WildClawBench: A Benchmark for Real-World, Long-Horizon Agent Evaluation

Large language and vision-language models increasingly power agents that act on a user's behalf through command-line interface (CLI) harnesses. However, most agent benchmarks still rely on synthetic sandboxes, short-horizon tasks, mock-service APIs, and final-answer checks, leaving open whether agents can complete realistic long-horizon work in the runtimes where they are deployed. This work presents WildClawBench, a native-runtime benchmark of 60 human-authored, bilingual, multimodal tasks spanning six thematic categories. Each task averages roughly 8 minutes of wall-clock time and over 20 tool calls, and runs inside a reproducible Docker container hosting an actual CLI agent harness (OpenClaw, Claude Code, Codex, or Hermes Agent) with access to real tools rather than mock services. Grading is hybrid, combining deterministic rule-based checks, environment-state auditing of side effects, and an LLM/VLM judge for semantic verification. Across 19 frontier models, the best, Claude Opus 4.7, reaches only 62.2% overall under OpenClaw, while every other model stays below 60%, and switching harness alone shifts a single model by up to 18 points. These results show that long-horizon, native-runtime agent evaluation remains a far-from-resolved task for current frontier models. We release the tasks, code, and containerized tooling to support reproducible evaluation.

cs.CL↗

arXiv2026

Visual-ERM: Reward Modeling for Visual Equivalence

Vision-to-code tasks require models to reconstruct structured visual inputs, such as charts, tables, and SVGs, into executable or structured representations with high visual fidelity. While recent Large Vision Language Models (LVLMs) achieve strong results via supervised fine-tuning, reinforcement learning remains challenging due to misaligned reward signals. Existing rewards either rely on textual rules or coarse visual embedding similarity, both of which fail to capture fine-grained visual discrepancies and are vulnerable to reward hacking. We propose Visual Equivalence Reward Model (Visual-ERM), a multimodal generative reward model that provides fine-grained, interpretable, and task-agnostic feedback to evaluate vision-to-code quality directly in the rendered visual space. Integrated into RL, Visual-ERM improves Qwen3-VL-8B-Instruct by +8.4 on chart-to-code and yields consistent gains on table and SVG parsing (+2.7, +4.1 on average), and further strengthens test-time scaling via reflection and revision. We also introduce VisualCritic-RewardBench (VC-RewardBench), a benchmark for judging fine-grained image-to-image discrepancies on structured visual data, where Visual-ERM at 8B decisively outperforms Qwen3-VL-235B-Instruct and approaches leading closed-source models. Our results suggest that fine-grained visual reward supervision is both necessary and sufficient for vision-to-code RL, regardless of task specificity.

cs.CV↗

arXiv2026

Exponential mixing for the stochastic Allen--Cahn equation with localized white noise

This paper studies the 1D stochastic Allen--Cahn equation on a bounded domain driven by localized white noise. We prove that the associated Markov process admits a unique invariant measure and is exponential mixing. The main challenge lies in the interaction between localized nature of the noise and non-trivial global dynamics of the system. To overcome this, our approach relies on two ingredients from PDE control theory: stabilization for the linearized system and global steady-state controllability for the nonlinear equation. The stabilization result is derived using the weak observability and Fenchel--Rockafellar duality, while the global controllability relies on quasi-static deformations combined with global dynamics.

math.PR↗

arXiv2026

qPRO-AQFP: Post-Routing Optimization of AQFP Circuits with Delay Line Clocking

Adiabatic Quantum-Flux-Parametron (AQFP) logic is an ultra-low-power superconducting logic family with energy consumption approaching the Shannon limit, making it attractive for quantum computing control and cryogenic computing systems. Traditional AQFP designs face significant physical design challenges due to strict gate-level clocking requirements and limited interconnect lengths, leading to substantial buffer overhead and difficult timing closure. Recently, delay-line clocking of AQFP has been proposed to improve timing margins and reduce latency by enabling more flexible clock scheduling. However, prior work has primarily focused on placement and latency minimization, while relying on fixed timing parameters that do not capture the frequency dependence of AQFP setup and hold constraints. To address this limitation, we propose a frequency-aware post-routing optimization framework that jointly optimizes clock period, latency, and timing slack under user-specified weighting. Experimental results across common benchmarks achieve 100% post-routing timing closure across a range of performance--latency--slack trade-offs. Our approach also automates phase-skipping, reducing path-balancing buffer insertion by 34% on average while only reducing operating frequency by 4%.

cs.ET↗

arXiv2026

Graph-to-Frame RAG: Visual-Space Knowledge Fusion for Training-Free and Auditable Video Reasoning

When video reasoning requires external knowledge, many systems with large multimodal models (LMMs) adopt retrieval augmentation to supply the missing context. Appending textual or multi-clip evidence, however, forces heterogeneous signals into a single attention space. We observe diluted attention and higher cognitive load even on non-long videos. The bottleneck is not only what to retrieve but how to represent and fuse external knowledge with the video backbone.We present Graph-to-Frame RAG (G2F-RAG), a training free and auditable paradigm that delivers knowledge in the visual space. On the offline stage, an agent builds a problem-agnostic video knowledge graph that integrates entities, events, spatial relations, and linked world knowledge. On the online stage, a hierarchical multi-agent controller decides whether external knowledge is needed, retrieves a minimal sufficient subgraph, and renders it as a single reasoning frame appended to the video. LMMs then perform joint reasoning in a unified visual domain. This design reduces cognitive load and leaves an explicit, inspectable evidence trail.G2F-RAG is plug-and-play across backbones and scales. It yields consistent gains on diverse public benchmarks, with larger improvements in knowledge-intensive settings. Ablations further confirm that knowledge representation and delivery matter. G2F-RAG reframes retrieval as visual space knowledge fusion for robust and interpretable video reasoning.

cs.CV↗

arXiv2026

Reinforce to Learn, Elect to Reason: A Dual Paradigm for Video Reasoning

Video reasoning has advanced with large multimodal models (LMMs), yet their inference is often a single pass that returns an answer without verifying whether the reasoning is evidence-aligned. We introduce Reinforce to Learn, Elect to Reason (RLER), a dual paradigm that decouples learning to produce evidence from obtaining a reliable answer. In RLER-Training, we optimize the policy with group-relative reinforcement learning (RL) and 3 novel task-driven rewards: Frame-sensitive reward grounds reasoning on explicit key frames, Think-transparency reward shapes readable and parsable reasoning traces, and Anti-repetition reward boosts information density. These signals teach the model to emit structured, machine-checkable evidence and potentiate reasoning capabilities. In RLER-Inference, we apply a train-free orchestrator that generates a small set of diverse candidates, parses their answers and cited frames, scores them by evidence consistency, confidence, transparency, and non-redundancy, and then performs a robust evidence-weighted election. This closes the loop between producing and using evidence, improving reliability and interpretability without enlarging the model. We comprehensively evaluate RLER against various open-source and RL-based LMMs on 8 representative benchmarks. RLER achieves state of the art across all benchmarks and delivers an average improvement of 6.3\% over base models, while using on average 3.1 candidates per question, indicating a favorable balance between compute and quality. The results support a simple thesis: making evidence explicit during learning and electing by evidence during inference is a robust path to trustworthy video reasoning.

cs.CV↗

arXiv2026

Development of a time calibration system for the KLM upgrade in the Belle II experiment

To meet the stringent time calibration requirements for the Belle II experiment upgrade, particularly for its large-size KL and Muon Detector comprising tens of thousands of scintillator channels with time resolutions better than 100 ps, we developed a compact and high-speed time calibration system. The system utilizes a laser diode as its light source, integrated with a fast pulse laser drive circuit that employs high-speed switching GaN FETs and gate drivers. A prototype was constructed and rigorously evaluated using scintillators, achieving timing resolutions of about 13 ps for a single calibration channel. Furthermore, internal deviations among calibration channels were analyzed, with most measurements remaining within 250 ps. These results highlight the system's precision, scalability, and suitability for large-scale particle physics experiments.

physics.ins-det↗

arXiv2026

On Sak's criterion for statistical models with long-range interaction

Determining the threshold value $σ_*$ that separates the short-range (SR) and long-range (LR) universality classes in phase transitions remains a controversial issue. While Sak's criterion, $σ_* = 2 - η_{\mathrm{SR}}$, has been widely accepted, recent studies of two-dimensional (2D) models with long-range interactions have challenged it. In this work, we focus on the crossover between LR and SR criticality in several classical 2D statistical models, including the XY, Heisenberg, percolation, and Ising models, whose interactions decay as $1/r^{2+σ}$. Our previous simulations for the XY, Heisenberg, and percolation models consistently indicate a universal boundary at $σ_* = 2$. Here, we complete the picture by performing large-scale Monte Carlo simulations of the 2D LR-Ising model, reaching lattice sizes up to $L = 8192$. By analyzing the Fortuin-Kasteleyn critical polynomial $R_p$, the Binder ratio $Q_m$, and the anomalous dimension $η$, we obtain convergent and self-consistent evidence that the universality class already changes sharply at $σ= 2$. Taken together, these results establish a unified scenario for LR interacting systems: across all studied models, the crossover from LR to SR universality occurs at $σ_* = 2$.

cond-mat.stat-mech↗

arXiv2025

LiveSecBench: A Dynamic and Event-Driven Safety Benchmark for Chinese Language Model Applications

We introduce LiveSecBench, a continuously updated safety benchmark specifically for Chinese-language LLM application scenarios. LiveSecBench constructs a high-quality and unique dataset through a pipeline that combines automated generation with human verification. By periodically releasing new versions to expand the dataset and update evaluation metrics, LiveSecBench provides a robust and up-to-date standard for AI safety. In this report, we introduce our second release v251215, which evaluates across five dimensions (Public Safety, Fairness & Bias, Privacy, Truthfulness, and Mental Health Safety.) We evaluate 57 representative LLMs using an ELO rating system, offering a leaderboard of the current state of Chinese LLM safety. The result is available at https://livesecbench.intokentech.cn/.

cs.CL↗

arXiv2025

ARM-Thinker: Reinforcing Multimodal Generative Reward Models with Agentic Tool Use and Visual Reasoning

Reward models are critical for aligning vision-language systems with human preferences, yet current approaches suffer from hallucination, weak visual grounding, and an inability to use tools for verification, limiting their reliability on complex multimodal reasoning tasks. We present ARM-Thinker, an A}gentic multimodal Reward Model that autonomously invokes external tools (e.g., image cropping, doc page retrieval) to ground judgments in verifiable evidence, replacing static, non-interactive reward scoring. This enables the model to verify fine-grained visual details, cross-reference multi-page evidence, and validate reasoning claims, which are capabilities absent in existing reward models. We train ARM-Thinker with multi-stage reinforcement learning, jointly optimizing tool-calling decisions and judgment accuracy. To evaluate agentic reward modeling, we introduce ARMBench-VL, comprising three benchmarks that assess fine-grained visual grounding (image-level tools), multi-page document understanding (retrieval tools), and instruction following (text-level verification). ARM-Thinker achieves +16.2% average improvement on reward modeling benchmarks, +9.6% on tool-use tasks, and outperforms baselines on multimodal math and logical reasoning benchmarks. Our results demonstrate that agentic capabilities significantly enhance both accuracy and interpretability of reward models.

cs.CV↗

arXiv2025

Asymptotic stability and mean ergodicity of Feller processes on Polish spaces

This article establishes several necessary and sufficient criteria on asymptotic stability and mean ergodicity in various types of topologies for Feller processes taking values in Polish spaces. In particular, asymptotic stability and mean ergodicity in Wasserstein distance and weighted total variation distance are considered. The characterizations are formulated by using the notions of generalized eventual continuity properties and lower bound conditions, where the proofs invoke the coupling approach.

math.PR↗

arXiv2025

A multifractal analysis for escaping trajectories on free groups

We consider a free group extension of a subshift of finite type $σ:Σ\rightarrowΣ$, and consider three sets of points in $Σ$ to which the corresponding trajectories on the free group escape to a given point in the Gromov boundary of the free group in three different senses. Under very mild conditions, we provide a common positive lower bound for the $u$-dimensions of these three sets. We also show that the lower bound is equal to the $u$-dimensions of these three sets when the extension and $u$ have certain symmetries. Moreover, we apply our results to geodesic flows on free group covers of Schottky surfaces, and show that there exists a dimension gap between the three sets we considered and the entire escaping set.

math.DS↗