Searcharxiv⌕ Search

arXiv subjects

Ruoyu Wang

Publications and source records attributed to Ruoyu Wang.

At least 37 records · Page 2Linked to original sources
arXiv2026

Statistical inference for the probability of necessity for causal attribution

To answer questions of "causes of effects", the probability of necessity was previously introduced for assessing whether an observed outcome was caused by an earlier treatment. However, statistical inference for the probability of necessity is understudied due to several difficulties, which hinder its application in practice. The evaluation of the probability of necessity involves the joint distribution of potential outcomes, and thus it is generally not point identified and one can at best obtain lower and upper bounds even in randomized experiments, unless fairly stringent monotonicity assumptions on potential outcomes are made. Moreover, these bounds are non-smooth functionals of the observed data distribution and standard estimation and inference methods cannot be directly applied. In this paper, we investigate the statistical inference for the probability of necessity in general situations where it may not be point identified. We introduce a mild margin condition to tackle the non-smoothness, under which the bounds become pathwise differentiable. We establish the semiparametric efficiency theory and propose novel asymptotically efficient estimators of the lower and upper bounds, and further construct confidence intervals for the probability of necessity based on the proposed bound estimators. The resultant confidence intervals can effectively utilize the observed covariates to reduce lengths. The proposed approach has potential application in biomedical, epidemiological, and legal studies where understanding causal attribution beyond traditional causal effects is essential.

stat.ME↗
arXiv2026

SWE-Review: Closing the Loop on Issue Resolution with Agentic Code Review

Coding agents increasingly generate pull requests (PRs) for real-world software issues, yet one-shot PR generation remains open-loop: the PR is proposed without systematic review, diagnosis, or revision. We introduce \textbf{SWE-Review}, a framework for closing this loop with agentic code review. Given an issue and an AI-generated PR, a reviewer agent explores the repository, decides whether the PR should be accepted, and provides structured feedback for revision. We evaluate this setting with our proposed \textbf{SWE-Review-Bench} to measure both review correctness and downstream revision usefulness. We further curate \textbf{SWE-Review-Traj} dataset to study broader applications of agentic review and fill the data-scarcity gap for open reviewer training. Experiments show that agentic review continuously improves PRs through a generate-review-revise loop, outperforms single-turn fixed-context review in both decision accuracy and resolve rate after revision, transfers beyond review to improve issue-resolution models, and enables effective and efficient test-time scaling. These results position agentic code review as a practical mechanism for moving AI coding agents from one-shot PR generation toward closed-loop issue resolution.

cs.SE↗
arXiv2026

Gradient-Flow Optimization as Dynamic Random-Effects Inference: Testing and Early Stopping with Applications to Deep Learning

Gradient-flow optimization is usually viewed as an algorithmic procedure for minimizing empirical loss, with training duration selected by validation or heuristic early stopping rules. We develop a statistical inference framework for gradient-flow training. We show that whenever fitted values evolve through a time-invariant positive semidefinite training operator, the output at each time is equivalent to the best linear unbiased predictor under a corresponding random-effects model. Training time then becomes a variance-component parameter governing variance reallocation from residual noise to structured signal. This turns two training decisions into inferential problems: whether training is needed becomes a variance-component test for signal beyond initialization, and how long to train becomes restricted maximum likelihood (REML) estimation of the training-time variance component. We show that the REML-guided early stopping rule selects the time at which optimized spectral losses become decorrelated from the training-operator eigenvalues. The asymptotic prediction optimality of the REML-guided early stopping time is established for fixed-design in-sample risk and random-design out-of-sample risk. Deep learning models in fixed-kernel gradient regimes provide canonical instantiations for our results. Numerical experiments and a UK Biobank proteomics application show competitive accuracy of the REML-guided early stopping time with reduced reliance on validation splits and repeated checkpoint evaluation.

stat.ML↗
arXiv2026

Adaptive and Efficient Learning with Blockwise Missing and Semi-Supervised Data

Data fusion enables powerful and generalizable analyses across multiple sources. However, different data collection capacities across different sources lead to blockwise missingness (BM), which poses challenges in practice. Meanwhile, the high cost of obtaining gold-standard labels leaves the majority of samples unlabeled, known as the semi-supervised (SS) problem. In this paper, we propose a novel Data-adaptive Estimation approach for data FUsion in the SEmi-supervised setting (DEFUSE) that handles both BM and SS issues in the presence of distributional shifts across data sources under a missing at random (MAR) mechanism}. DEFUSE starts with a complete-data-only estimator derived from the primary data source, and uses data-adaptive and distributional-shift-adjusted procedures to successively incorporate the data with BM covariates and the large unlabeled sample to effectively reduce the estimation variance without incurring bias. To further avoid bias due to fusion of misaligned data violating of the MAR assumption, a screening method is developed to identify and exclude data sources that are not aligned with the primary source. Compared to existing approaches, DEFUSE offers two main improvements. First, it offers a new data-adaptive control variate approach to handle BM, which achieves intrinsic efficiency and robustness against distributional shifts. Second, it reveals a more essential role for the unlabeled sample in the BM regression problem, leading to improved estimation. These advantages are theoretically guaranteed and empirically supported by simulation studies and two real-world biomedical applications.

stat.ME↗
arXiv2026

SparseWorld: Enhancing End-to-End Autonomous Driving via World Models with Sparse Scene Representation

Recently, world models have made significant progress in enhancing end-to-end driving systems through both future situation forecasting and improved scene understanding. However, existing driving world models are typically built upon dense scene representations, causing high computational costs and redundant information. In this paper, we present SparseWorld, a lightweight world model that focuses on predicting only the critical layout of the scene, enabling efficient future forecasting for end-to-end driving systems. SparseWorld first performs autoregressive rollout to forecast future map elements and surrounding agents, enabling the model to learn how driving scenarios evolve over time. It then leverages these predicted futures to refine downstream motion prediction and trajectory planning. Specifically, we propose a Sparse Dreamer that anticipates future instances in the latent space through joint temporal and spatial attention. By interacting with predicted future instances, the motion planner captures more accurate motion patterns and generates more informed and safety-aware trajectories. Extensive experiments demonstrate that SparseWorld significantly reduces collision risk and achieves state-of-the-art performance on the open-loop planning metrics of the nuScenes dataset with a collision rate of 0.05\%. Moreover, it substantially outperforms the baseline method in closed-loop planning metrics on the Bench2Drive benchmark. Supplementary material is available at the project page: https://wryzju.github.io/SparseWorld/.

cs.CV↗
arXiv2026

ARVO: Atlas of Reproducible Vulnerabilities for Open-Source Software

Achieving reproducibility, quantity, and diversity in vulnerability datasets has long been viewed as an inherent three-way trade-off, where improving one dimension often comes at the cost of the others. In practice, reproducibility has been the dimension most often neglected. This has limited what can be automatically extracted from historical bug datasets, and has reduced their utility for downstream security research. In this work, we propose a method to produce a new security dataset which ensures reproducibility for diverse vulnerabilities at scale by identifying the key obstacles to large-scale bug reproduction and addressing them with general solutions. Using this method, we introduce full reproducibility to the largest open source software vulnerability dataset (OSS-Fuzz) and construct the ARVO dataset (an Atlas of Reproducible Vulnerabilities in Open-source software). ARVO is a large-scale dataset consisting of over 6,100 real-world vulnerabilities across 311 projects. Focusing on reproducibility, ARVO differs from existing datasets by providing each vulnerability in a form that can be consistently rebuilt, triggered, and analyzed across versions. Reproducibility also enables automatic identification of the corresponding patch for each vulnerability and supports direct interaction with vulnerabilities after code changes, capabilities that existing large-scale datasets do not provide. In our evaluation, ARVO successfully reproduces 81% of vulnerabilities and achieves 89.4% accuracy on the located patches. We also discuss ARVO's influence on both upstream practices and downstream security research.

cs.CR↗
arXiv2026

ARVO: Atlas of Reproducible Vulnerabilities for Open-Source Software

Achieving reproducibility, quantity, and diversity in vulnerability datasets has long been viewed as an inherent three-way trade-off, where improving one dimension often comes at the cost of the others. In practice, reproducibility has been the dimension most often neglected. This has limited what can be automatically extracted from historical bug datasets, and has reduced their utility for downstream security research. In this work, we propose a method to produce a new security dataset which ensures reproducibility for diverse vulnerabilities at scale by identifying the key obstacles to large-scale bug reproduction and addressing them with general solutions. Using this method, we introduce full reproducibility to the largest open source software vulnerability dataset (OSS-Fuzz) and construct the ARVO dataset (an Atlas of Reproducible Vulnerabilities in Open-source software). ARVO is a large-scale dataset consisting of over 6,100 real-world vulnerabilities across 311 projects. Focusing on reproducibility, ARVO differs from existing datasets by providing each vulnerability in a form that can be consistently rebuilt, triggered, and analyzed across versions. Reproducibility also enables automatic identification of the corresponding patch for each vulnerability and supports direct interaction with vulnerabilities after code changes, capabilities that existing large-scale datasets do not provide. In our evaluation, ARVO successfully reproduces 81% of vulnerabilities and achieves 89.4% accuracy on the located patches. We also discuss ARVO's influence on both upstream practices and downstream security research.

cs.CR↗
arXiv2026

Semiparametric Mediation Analysis with Separately Observed Mediator and Outcome under Unmeasured Confounding

Mediation analysis is widely used to disentangle causal pathways, yet in many real-world studies the mediator M and outcome Y are never jointly observed. This incompleteness breaks the standard identification strategy for natural direct and indirect effects. We introduce a novel data fusion framework that restores the identification by combining two incomplete data sources, one measuring $M$ and the other measuring Y. Our approach leverages shared instrumental variables (IVs) to circumvent the need to observe (M,Y) jointly, remains valid under unmeasured confounding via a no-interaction condition, and accommodates covariate and exposure shifts across data sources under a latent alignment condition. We establish two identification strategies, one for settings with a known set of valid IVs, and another for settings where valid IVs must be learned. We further develop semiparametric, influence-function-based estimators with multiple robustness properties, and propose an estimator that attains the semiparametric efficiency bound under appropriate conditions. We apply our framework to quantify the extent to which the effect of SNP rs610932 on dementia risk is mediated through immune-related gene-expression pathways.

stat.ME↗
arXiv2026

Exploring the Design Space of Reward Backpropagation for Flow Matching

Aligning text-to-image flow matching models with human preferences via direct reward backpropagation is sample-efficient but hampered by two well-known pathologies: activations cannot be stored across the full sampling trajectory at modern model scale, and chained Jacobian products across steps inflate the reward gradient as it travels back to early indices. Connector-based methods, such as LeapAlign, address these issues by replacing the full backward trajectory with a short pinned path, highlighting a useful decoupling between sampling and optimization. However, the quality of the resulting gradient depends on how accurately this short path approximates the full rollout, especially over long intervals. We propose FlowBP, a unified surrogate-trajectory framework that treats the backward trajectory itself as the design object. FlowBP keeps a no-gradient cached rollout for sampling, then builds a lightweight backward surrogate from cached and selectively re-forwarded velocities. This view separates four choices: the reward-model input, active set, integration weights, and bridge coupling, and recovers prior direct-gradient methods as particular settings. Within this framework, we instantiate three variants: FlowBP-Sparse uses sparse Euler reconstruction, FlowBP-Bridge adds controlled bridge coupling, and FlowBP-Lagrange raises the order of leap quadrature. All three bound memory by the active-set size and limit gradient chaining to at most one Jacobian factor. Across SD3.5-M, FLUX.1-dev, and FLUX.2-Klein-base on preference, quality, and compositional metrics, the three variants improve over direct-gradient baselines on most metrics.

cs.LG↗
arXiv2026

Divide-and-shrink: An efficient and heterogeneity-agnostic approach for transfer estimation using summary statistics

Knowledge transfer across data sources holds great promise for improving the estimation of target population parameters by leveraging the growing availability of data from different sources. However, the effectiveness of knowledge transfer is often challenged by the complex and pervasive heterogeneity between data sources and the lack of access to individual-level data. This paper proposes the divide-and-shrink (dShrink) method, a transfer estimation method that estimates target population parameters in a closed form using summary statistics from a target population and some external source populations while accounting for population heterogeneity. The dShrink estimator is guaranteed to outperform the estimator based solely on the target population in terms of expected quadratic error under arbitrary population heterogeneity. The gain can be substantial when the target and source populations are similar, or the underlying true parameter values are near zero. Notably, dShrink is model-free, requires no user-specified tuning parameters, robust to various types of heterogeneity between data sources, and applies to a broad range of parameter estimation problems. dShrink remains effective even when the covariance matrix is not accessible for the external summary statistics and offers flexibility in incorporating side information and summary statistics from multiple source populations. Simulations and real data analyses demonstrate the superior performance of the dShrink estimator and its potential as a robust tool for transfer estimation.

stat.ME↗
arXiv2026

Reinforcing Few-step Generators via Reward-Tilted Distribution Matching

Recent advances in few-step diffusion distillation have enabled efficient image generation, yet aligning these models with human preferences remains challenging. We propose Reward-Tilted Distribution Matching Distillation (RTDMD), a two-stage framework that unifies distribution matching distillation with reward-guided reinforcement learning for few-step flow generators. We show that minimizing the KL divergence to a reward-tilted teacher distribution naturally decomposes into a distribution matching term and a reward maximization term. In the first stage, we introduce Ambient-Consistent Distribution Matching Distillation (AC-DMD), which performs subinterval-wise distribution matching and augments the fake score objective with a consistency regularizer to help the fake score model track the shifting generator distribution under limited updates. In the second stage, we jointly optimize both terms: for the reward maximization term, we derive a hybrid policy gradient that combines a GRPO-style estimator for the stochastic intermediate transitions with direct reward backpropagation through the deterministic final step, and further introduce step-subset GRPO (SubGRPO) to reduce variance. Experiments on SD3, SD3.5, and FLUX.2 demonstrate that RTDMD establishes new state-of-the-art results across preference, aesthetic, and compositional metrics with only 4 inference steps, outperforming previous few-step text-to-image generation methods. Code and models are available at https://github.com/Harahan/RTDMD.

cs.CV↗
arXiv2026

What Makes Interaction Trajectories Effective for Training Terminal Agents?

Stronger code agents are commonly assumed to be superior teachers for post-training, yet this assumption remains poorly disentangled from task difficulty, harness design, and student capacity. We investigate this pedagogical link using Terminal-Lego, a scalable pipeline that transforms multi-domain real-world issues into environment-verified agentic tasks. Surprisingly, standalone performance does not dictate teaching efficacy: while Claude Opus 4.6 achieves higher scores on Terminal-Bench 2.0, students fine-tuned on trajectories from DeepSeek-V3.2, a lower-scoring agent, exhibit significantly stronger generalization. We attribute this "pedagogical paradox" to Environment-Grounded Supervision (EGS): trajectories that explicitly expose inspect-act-verify behaviors through harness-visible interactions allow students to internalize robust problem-solving routines rather than fragile action sequences. Scaling analysis reveals exceptional data efficiency: with only 15.3k Terminal-Lego trajectories, for example, Qwen3-32B achieves a 24.3% score on Terminal-Bench 2.0, rivaling previous SOTA performance established with over 30x the data volume. Our results suggest that the frontier of agent post-training lies beyond mere outcome-matching, shifting the focus toward "Harness Engineering", where the systematic design of environment-grounded interaction structures serves as the primary catalyst for reproducible and generalizable agentic intelligence.

cs.AI↗
arXiv2026

Restoring Initial Noise Sensitivity in Text-to-Image Distillation via Geometric Alignment

Generative distillation significantly accelerates text-to-image (T2I) generation by compressing multi-step trajectories into few-step student models while preserving perceptual quality. However, existing methods primarily optimize efficiency and output fidelity, often neglecting critical properties of the original trajectory. In this work, we identify a key missing property: sensitivity to initial noise, whose degradation impairs downstream control methods relying on noise-based optimization and manipulation. We trace this issue to standard distillation objectives that enforce pointwise output alignment, inadvertently flattening the input-output landscape and suppressing the teacher's local geometric structure. To address this, we propose Geometry-Aware Distillation (GAD), a sensitivity-preserving framework that aligns the local functional behavior of teacher and student models. Specifically, GAD matches Jacobian-vector products with respect to input noise, enabling the student to reproduce the teacher's differential response to perturbations. Extensive experiments across multiple T2I paradigms and noise-driven control tasks demonstrate that GAD significantly restores sensitivity and improves diversity while maintaining high visual fidelity. Code is available at https://github.com/Hannah1102/GAD.

cs.CV↗
arXiv2026

Benjamini-Schramm convergence and subtrees of trees

In this paper, we study the asymptotic behaviour of the number of subtrees and the subtree density for a sequence of trees that converges in the Benjamini-Schramm sense. Benjamini-Schramm convergence, also called local weak convergence, describes the local behaviour of a sequence of graphs. Here we show that for a Benjamini-Schramm-convergent sequence of trees, the subtree entropy, i.e., the logarithm of the number of subtrees divided by the order, converges to a constant depending only on the limit. The same holds true for the subtree density, i.e., the probability of a uniformly random vertex being contained in a uniformly random subtree, provided that long paths are ruled out in the limit. Related to this, we show that the subtree density and the average subtree entropy are dense in different parts of the unit interval $[0,1]$ for both general trees and series-reduced trees.

math.CO↗
arXiv2026

SOWing Information: Cultivating Contextual Coherence with MLLMs in Image Generation

Originating from the diffusion phenomenon in physics, which describes the random movement and collisions of particles, diffusion generative models simulate a random walk in the data space along the denoising trajectory. This allows information to diffuse across regions, yielding harmonious outcomes. However, the chaotic and disordered nature of information diffusion in diffusion models often results in undesired interference between image regions, causing degraded detail preservation and contextual inconsistency. In this work, we address these challenges by reframing disordered diffusion as a powerful tool for text-vision-to-image generation (TV2I) tasks, achieving pixel-level condition fidelity while maintaining visual and semantic coherence throughout the image. We first introduce Cyclic One-Way Diffusion (COW), which provides an efficient unidirectional diffusion framework for precise information transfer while minimizing disruptive interference. Building on COW, we further propose Selective One-Way Diffusion (SOW), which utilizes Multimodal Large Language Models (MLLMs) to clarify the semantic and spatial relationships within the image. Based on these insights, SOW combines attention mechanisms to dynamically regulate the direction and intensity of diffusion according to contextual relationships. Extensive experiments demonstrate the untapped potential of controlled information diffusion, offering a path to more adaptive and versatile generative models in a learning-free manner.

cs.CV↗
arXiv2026

The Distribution Of Subtrees In Dense Graphs And The Roots Of The Subtree Polynomial

For a graph $G$ with $n$ vertices and a positive integer $k \leq n$, let $s_k(G)$ be the number of subtrees (subgraphs that are trees, not necessarily induced) of $G$ with $k$ vertices. The subtree polynomial of $G$ is $S(G;x) = \sum_{k=1}^n s_k(G) x^k$. In this paper, we consider dense connected graphs with a minimum degree that is linear in the number of vertices. We prove that the number of missing vertices in a random subtree is asymptotically Poisson-distributed and deduce that all the roots of the subtree polynomial have to be close to $0$.

math.CO↗
arXiv2026

Root-Cause-Driven Automated Vulnerability Repair

Recent LLM-based systems have made automated vulnerability repair increasingly practical, but two challenges remain. First, without strong signals about where a bug originates, repair agents drift toward shallow edits that silence the observed failure while leaving the underlying defect unresolved. Second, finding the root cause for bugs is hard: even developers familiar with the codebase frequently produce fixes that address symptoms rather than the root cause, and LLM-based agents, operating with noisier context and less program understanding, are no exception. We present Kumushi, a root-cause-driven patching agent that addresses both challenges by combining diversified dynamic fault localization with evidence-weighted ranking to focus the LLM on the code most relevant to the defect. To rigorously measure whether Kumushi produces genuinely better patches, we also introduce a two-tier patch quality metric that pairs automated oracle validation with structured expert assessment of patches. Evaluated on 178 C/C++ vulnerabilities, Kumushi substantially outperforms prior specialized repair agents under automated evaluation while matching a frontier commercial coding agent. Expert assessment then reveals differences that oracles cannot: Kumushi produces more root-cause fixes and fewer superficial patches, and is preferred in the majority of decisive pairwise comparisons. Together, these results demonstrate that progress in automated vulnerability repair requires not only stronger patching systems, but also richer evaluation methods capable of distinguishing genuine fixes from oracle-passing ones.

cs.CR↗
arXiv2026

SCRIBE: Practical Static Binary Patching via Binary-Aware Recompilation of Decompiled Code

When source code or the original toolchain is unavailable, patching binaries is difficult because it requires editing low-level assembly code directly. As an alternative, one can decompile the binary, apply the patch at the source level, and then recompile the modified code. However, as this paper demonstrates, this workflow is hindered by pervasive syntactic and semantic inaccuracies in the output of modern decompilers, many of which prior work has overlooked. To address these challenges, we present SCRIBE, a patching framework that handles syntactic and semantic issues in decompiled code, improving both recompilation success and correctness. SCRIBE's novel "binary-aware" recompilation approach repairs semantic inaccuracies in decompiler output by leveraging information extracted directly from the original binary. In our evaluation, SCRIBE resolved approximately 81% of previously incorrect functions produced by the Hex-Rays decompiler, demonstrating the effectiveness of its approach. Moreover, we show that, using SCRIBE, it is possible to patch 13 of 14 real-world CVEs without access to the original source code and without performing any manual binary editing. To further validate our findings, we conducted a user study with 18 participants. Using SCRIBE, participants achieved 100% patching success, compared to 3.7% without it. Finally, we asked three large language models to generate source-level patches via SCRIBE; all three achieved 100% success when using the framework, demonstrating its potential to enable fully automated patching. Overall, these results indicate that SCRIBE makes source-level patching of binaries accessible and reliable, even without access to the original source.

cs.CR↗