arXiv subjects

Peixin Wang

Publications and source records attributed to Peixin Wang.

At least 19 recordsLinked to original sources

arXiv2026

CAFE: Self-Improving Search Agents Need Co-Evolving Feedback

Reliable search requires more than acquiring external evidence. An agent must also recognize and recover from errors as its trajectory unfolds. In-trajectory feedback provides a mechanism for such recovery by diagnosing where the search has drifted and redirecting subsequent reasoning steps. This is particularly important in long-horizon search, where an early directional error may receive no immediate corrective signal and can compound across later steps. Making such feedback learnable, however, creates a coupled problem: the agent must learn when to request and use feedback, while the critic must learn corrections from outcome-confounded rollouts as the agent's failure patterns evolve. We introduce CAFE (Coupled Agent--Feedback Evolution), a framework in which a shared-parameter model alternates between search-agent and critic roles. CAFE initializes feedback-conditioned recovery from trajectories built around the base agent's own failures, then couples online and offline optimization. During online RL, a comparative feedback estimate uses a prompt-level call--skip success gap to shape request returns, while feedback-aware advantage shaping reweights token advantages before and after feedback. Offline, rollout-derived preference optimization learns feedback from matched successful and unsuccessful trajectories. On seven agentic search benchmarks, CAFE outperforms the evaluated RL-based search agents on average, retains its gains across all six out-of-domain benchmarks, and reduces answer-level hallucinations. One-sided ablations show that improving only the agent or only the critic eventually plateaus, whereas alternating the two updates continues to improve performance. These findings suggest that a self-improving search agent needs feedback that co-evolves with the policy it guides.

cs.AI↗

arXiv2026

EchoFuzz: Empowering Smart Contract Fuzzing with Large Language Models

Smart contracts, serving as the cornerstone of decentralized applications, autonomously manage trillion-dollar digital assets, making them attractive targets for attacks. Fuzzing has emerged as a promising technique for detecting vulnerabilities in smart contracts, yet existing methods face two main challenges. (1) The logical gap in state transitions and combinatorial redundancy hinders effective tradeoffs between bug detection efficiency and state space exploration cost, leading to critical execution paths to be overlooked. (2) Rule-based sequence mutation strategies suffer from path redundancy and inadequate guidance from contract logic, resulting in performance bottlenecks that stall the exploration of in-depth vulnerability-oriented paths. To tackle these challenges, we propose EchoFuzz, an LLM-guided fuzzing framework introducing Vulnerable Function Call Sequences (VFCS) - minimal, behavior-preserving execution paths that expose bugs through key state transitions. EchoFuzz consists of two key procedures. First, we develop a chain-guided LLM approach, that combines static analysis with logical understanding to generate contract-specific VFCS candidates that eliminate combinatorial redundancy. Second, we adopt an iterative fuzzing strategy that uses LLMs with real-time feedback to adaptively steer fuzzer toward uncovered branches. Experiments show EchoFuzz outperforms state-of-the-art methods, achieving 29\% higher branch coverage and detecting 62\% more vulnerabilities. It also found 37 previously unknown vulnerabilities in real contracts, showing strong practicality.

cs.SE↗

arXiv2026

Mitigating Object Hallucinations in LVLMs via Attention Imbalance Rectification

Object hallucination in Large Vision-Language Models (LVLMs) severely compromises their reliability in real-world applications, posing a critical barrier to their deployment in high-stakes scenarios such as autonomous driving and medical image analysis. Through systematic empirical investigation, we identify that the imbalanced attention allocation, both across modalities (i.e., vision and language) and within modalities (among individual tokens), exhibits a strong causal correlation with the occurrence of object hallucination. Leveraging this insight, we introduce a novel concept termed attention imbalance, which not only quantifies the degree of attention disparity but also visually delineates the underlying patterns (e.g., over-attentiveness to irrelevant language tokens or under-attentiveness to discriminative visual features) that drive object hallucination. To mitigate object hallucination, we further propose Attention Imbalance Rectification (AIR), a lightweight decoding-time intervention method that reallocates attention weights and adjusts attention distributions to rectify modality-wise and token-wise imbalances. Extensive evaluations on four mainstream LVLMs and three benchmarks (CHAIR, POPE, and MM-Vet) with seven baselines demonstrate that AIR consistently reduces object hallucination rates, achieving up to a 35.1% reduction compared to the baselines, while improving up to 15.9% of LVLMs' general capability across diverse vision-language tasks.

cs.CV↗

arXiv2026

Entropy Is Not Enough: Unlocking Effective Reinforcement Learning for Visual Reasoning via Vision-Anchored Token Selection

While token-level entropy is commonly recognized as effective for credit assignment in text-only reinforcement learning with verifiable rewards (RLVR), it remains unclear whether this mechanism still holds in visual reasoning. Our controlled study shows that this mechanism collapses in visual reasoning due to the omission of vision-sensitive tokens with naturally low entropy. Although existing multimodal RL methods increasingly acknowledge the importance of visual perception, they struggle to satisfy the inherent demand for interleaving precise perceptual grounding with semantic reasoning, either lacking systematic visual measurements or overlooking that token entropy primarily drives semantic exploration. To address this, we introduce VEPO (Vision-Entropy token-selection for Policy Optimization), an effective RL framework explicitly integrating visual sensitivity with token entropy via a principled multiplicative coupling, where VEPO redirects gradient credit toward tokens which are simultaneously visually grounded and highly informative. Extensive experiments demonstrate VEPO's leading performance, significantly outperforming the entropy-only baseline by 2.28 points at 7B-scale and 3.15 points at 3B-scale. Ablations further substantiate the soundness of our method.

cs.AI↗

arXiv2026

Refined Barrier Conditions for Finite-Time Safety and Reach-Avoid Guarantees in Stochastic Systems

Providing finite-time probabilistic safety and reach-avoid guarantees is crucial for safety-critical stochastic systems. Existing state-of-the-art barrier methods often rely on a restrictive boundedness assumption for auxiliary functions, limiting their applicability. This paper presents refined barrier conditions that remove this assumption. Specifically, we establish conditions for deriving upper bounds on finite-time safety probabilities in discrete-time systems and lower bounds on finite-time reach-avoid probabilities in continuous-time systems. This relaxation expands the class of verifiable systems, especially those with unbounded state spaces, and facilitates the use of advanced optimization techniques, such as semi-definite programming with polynomial functions. Numerical examples demonstrate the effectiveness of the approach.

eess.SY↗

arXiv2026

Quantitative Verification of Constrained Occupation Time for Stochastic Discrete-time Systems

This paper addresses the quantitative verification of constrained occupation time in stochastic discrete-time systems, focusing on the probability of visiting a target set at least $k$ times while maintaining safety. Such cumulative properties are essential for certifying repeated behaviors like surveillance and periodic charging. To address this, we present the first barrier certificate framework capable of certifying these behaviors. We introduce multiplicative stochastic barrier functions that encode visitation counts implicitly within the algebraic structure of a scalar barrier. By adopting a switched-system reformulation to handle safety, we derive rigorous probabilistic bounds for both finite and infinite horizons. Specifically, we show that dissipative barriers establish upper bounds ensuring the exponential decay of frequent visits, while attractive barriers provide lower bounds via submartingale analysis. The efficacy of the proposed framework is demonstrated through numerical examples.

eess.SY↗

arXiv2026

A Formal Framework for Predicting Distributed System Performance under Faults (Extended Version)

Today's distributed systems operate in complex environments that inevitably involve faults and even adversarial behaviors. Predicting their performance under such environments directly from formal designs remains a longstanding challenge. We present the first formal framework that systematically enables performance prediction of distributed systems across diverse faulty scenarios. Our framework features a fault injector together with a wide range of faults, reusable as a library, and model compositions that integrate the system and the fault injector into a unified model suitable for statistical analysis of performance properties such as throughput and latency. We formalize the framework in Maude and implement it as an automated tool, PERF. Applied to representative distributed systems, PERF accurately predicts system performance under varying fault settings, with estimations from formal designs consistent with evaluations on real deployments.

cs.DC↗

arXiv2026

Neural Theorem Proving for Verification Conditions: A Real-World Benchmark

Theorem proving is fundamental to program verification, where the automated proof of Verification Conditions (VCs) remains a primary bottleneck. Real-world program verification frequently encounters hard VCs that existing Automated Theorem Provers (ATPs) cannot prove, leading to a critical need for extensive manual proofs that burden practical application. While Neural Theorem Proving (NTP) has achieved significant success in mathematical competitions, demonstrating the potential of machine learning approaches to formal reasoning, its application to program verification--particularly VC proving--remains largely unexplored. Despite existing work on annotation synthesis and verification-related theorem proving, no benchmark has specifically targeted this fundamental bottleneck: automated VC proving. This work introduces Neural Theorem Proving for Verification Conditions (NTP4VC), presenting the first real-world multi-language benchmark for this task. From real-world projects such as Linux and Contiki-OS kernel, our benchmark leverages industrial pipelines (Why3 and Frama-C) to generate semantically equivalent test cases across formal languages of Isabelle, Lean, and Rocq. We evaluate large language models (LLMs), both general-purpose and those fine-tuned for theorem proving, on NTP4VC. Results indicate that although LLMs show promise in VC proving, significant challenges remain for program verification, highlighting a large gap and opportunity for future research.

cs.AI↗

arXiv2025

Quantitative Verification of Omega-regular Properties in Probabilistic Programming

Probabilistic programming provides a high-level framework for specifying statistical models as executable programs with built-in randomness and conditioning. Existing inference techniques, however, typically compute posterior distributions over program states at fixed time points, most often at termination, thereby failing to capture the temporal evolution of probabilistic behaviors. We introduce temporal posterior inference (TPI), a new framework that unifies probabilistic programming with temporal logic by computing posterior distributions over execution traces that satisfy omega-regular specifications, conditioned on possibly temporal observations. To obtain rigorous quantitative guarantees, we develop a new method for computing upper and lower bounds on the satisfaction probabilities of omega-regular properties. Our approach decomposes Rabin acceptance conditions into persistence and recurrence components and constructs stochastic barrier certificates that soundly bound each component. We implement our approach in a prototype tool, TPInfer, and evaluate it on a suite of benchmarks, demonstrating effective and efficient inference over rich temporal properties in probabilistic models.

cs.PL↗

arXiv2025

A Minimalist Proof Language for Neural Theorem Proving over Isabelle/HOL

Neural Theorem Proving (NTP) employs LLMs to automate formal proofs in proof assistants. While LLMs have achieved relatively remarkable success in informal reasoning tasks using natural languages, the transition to mechanized formal theorem proving presents persistent challenges. Mechanized proof languages often contain many syntactic constructs and diverse, specialized proof tactics, which facilitate expert use but have no direct counterpart in informal mathematical proofs. These prover-specific idioms represent an additional burden for LLM-based NTPs that might be otherwise successful in generating informal proofs. Seeking to bridge this gap between formal proof construction and informal reasoning, in order to better facilitate NTP, this work approaches these challenges from a language design perspective. We look at common reasoning patterns in informal proofs and in existing mechanized proofs, and design Minilang -- a minimalist proof language that captures these reasoning patterns. In contrast to proof languages (informal and formal) that often feature a large collection of operations with unclear semantic boundaries, Minilang is deliberately kept minimalist -- its core design comprises only 10 operations, each with clear semantic distinctions. We further develop a rule-based translator from Isabelle's language (Isar) to Minilang, translating ~340K existing proofs with an ~85% success rate. Using this translated corpus, we finetune two LLMs to compare machine learning performance on Minilang versus the original Isar. Experiments show Minilang benefits the two LLMs by improving the pass@1 success rate on the PISA benchmark by up to 20/29 percentage points in comparison to the Isar-based LLMs w/wo Sledgehammer. The pass@1 rate reaches 69.1%, exceeding the prior work Baldur's pass@64 (65.7%); the pass@8 rate reaches 79.2%, exceeding the SOTA on PISA (71.0%) achieved by Magnushammer.

cs.PL↗

arXiv2025

Comparative Analysis of Barrier-like Function Methods for Reach-Avoid Verification in Stochastic Discrete-Time Systems

In this paper, we compare several representative barrier-like conditions from the literature for infinite-horizon reach-avoid verification of stochastic discrete-time systems. Our comparison examines both their theoretical properties and computational tractability, highlighting each condition's strengths and limitations that affect applicability and conservativeness. Finally, we illustrate their practical performance through computational experiments using semidefinite programming (SDP) and counterexample-guided inductive synthesis (CEGIS).

eess.SY↗

arXiv2025

Nex-N1: Agentic Models Trained via a Unified Ecosystem for Large-Scale Environment Construction

The evolution of Large Language Models (LLMs) from passive responders to autonomous agents necessitates a fundamental shift in learning paradigms -- from static imitation to incentive-driven decision making. However, this transition is significantly impeded by the lack of scalable infrastructure capable of constructing high-quality interaction signals for effective policy learning. To address this, we introduce a comprehensive method designed to systematically scale the diversity and complexity of interactive environments. Our method realizes this scaling by addressing three orthogonal dimensions: (1) Complexity: NexAU, a flexible agent framework that supports building complex agent hierarchies via simple configurations; (2) Diversity: NexA4A automatically generates diverse agent hierarchies from natural language to cover infinite domains; and (3) Fidelity: NexGAP bridges the simulation-reality gap by integrating dynamic real-world environment for grounded trajectories synthesis. We train Nex-N1 upon the diverse and complex interactive environments established by our infrastructure. Empirical results on benchmarks such as SWE-bench and tau2 demonstrate that Nex-N1 consistently outperforms SOTA open-source models and achieves competitive performance against frontier proprietary models on complex agentic tasks. We open-source the Nex ecosystem and model weights to facilitate further research.

cs.CL↗

arXiv2025

Structural Abstraction and Refinement for Probabilistic Programs

In this paper, we present structural abstraction refinement, a novel framework for verifying the threshold problem of probabilistic programs. Our approach represents the structure of a Probabilistic Control-Flow Automaton (PCFA) as a Markov Decision Process (MDP) by abstracting away statement semantics. The maximum reachability of the MDP naturally provides a proper upper bound of the violation probability, termed the structural upper bound. This introduces a fresh ``structural'' characterization of the relationship between PCFA and MDP, contrasting with the traditional ``semantical'' view, where the MDP reflects semantics. The method uniquely features a clean separation of concerns between probability and computational semantics that the abstraction focuses solely on probabilistic computation and the refinement handles only the semantics aspect, where the latter allows non-random program verification techniques to be employed without modification. Building upon this feature, we propose a general counterexample-guided abstraction refinement (CEGAR) framework, capable of leveraging established non-probabilistic techniques for probabilistic verification. We explore its instantiations using trace abstraction. Our method was evaluated on a diverse set of examples against state-of-the-art tools, and the experimental results highlight its versatility and ability to handle more flexible structures swiftly.

cs.FL↗

arXiv2024

Static Posterior Inference of Bayesian Probabilistic Programming via Polynomial Solving

In Bayesian probabilistic programming, a central problem is to estimate the normalised posterior distribution (NPD) of a probabilistic program with conditioning via score (a.k.a. observe) statements. Most previous approaches address this problem by Markov Chain Monte Carlo and variational inference, and therefore could not generate guaranteed outcomes within a finite time limit. Moreover, existing methods for exact inference either impose syntactic restrictions or cannot guarantee successful inference in general. In this work, we propose a novel automated approach to derive guaranteed bounds for NPD via polynomial solving. We first establish a fixed-point theorem for the wide class of score-at-end Bayesian probabilistic programs that terminate almost-surely and have a single bounded score statement at program termination. Then, we propose a multiplicative variant of Optional Stopping Theorem (OST) to address score-recursive Bayesian programs where score statements with weights greater than one could appear inside a loop. Finally, we use polynomial solving to implement our fixed-point theorem and OST variant. To improve the accuracy of the polynomial solving, we further propose a truncation operation and the synthesis of multiple bounds over various program inputs. Our approach can handle Bayesian probabilistic programs with unbounded while loops and continuous distributions with infinite supports. Experiments over a wide range of benchmarks show that compared with the most relevant approach (Beutner et al., PLDI 2022) for guaranteed NPD analysis via recursion unrolling, our approach is more time efficient and derives comparable or even tighter NPD bounds. Furthermore, our approach can handle score-recursive programs which previous approaches could not.

cs.PL↗

arXiv2024

Unifying Qualitative and Quantitative Safety Verification of DNN-Controlled Systems

The rapid advance of deep reinforcement learning techniques enables the oversight of safety-critical systems through the utilization of Deep Neural Networks (DNNs). This underscores the pressing need to promptly establish certified safety guarantees for such DNN-controlled systems. Most of the existing verification approaches rely on qualitative approaches, predominantly employing reachability analysis. However, qualitative verification proves inadequate for DNN-controlled systems as their behaviors exhibit stochastic tendencies when operating in open and adversarial environments. In this paper, we propose a novel framework for unifying both qualitative and quantitative safety verification problems of DNN-controlled systems. This is achieved by formulating the verification tasks as the synthesis of valid neural barrier certificates (NBCs). Initially, the framework seeks to establish almost-sure safety guarantees through qualitative verification. In cases where qualitative verification fails, our quantitative verification method is invoked, yielding precise lower and upper bounds on probabilistic safety across both infinite and finite time horizons. To facilitate the synthesis of NBCs, we introduce their $k$-inductive variants. We also devise a simulation-guided approach for training NBCs, aiming to achieve tightness in computing precise certified lower and upper bounds. We prototype our approach into a tool called $\textsf{UniQQ}$ and showcase its efficacy on four classic DNN-controlled systems.

cs.LG↗

arXiv2023

Robustness Verification of Deep Reinforcement Learning Based Control Systems using Reward Martingales

Deep Reinforcement Learning (DRL) has gained prominence as an effective approach for control systems. However, its practical deployment is impeded by state perturbations that can severely impact system performance. Addressing this critical challenge requires robustness verification about system performance, which involves tackling two quantitative questions: (i) how to establish guaranteed bounds for expected cumulative rewards, and (ii) how to determine tail bounds for cumulative rewards. In this work, we present the first approach for robustness verification of DRL-based control systems by introducing reward martingales, which offer a rigorous mathematical foundation to characterize the impact of state perturbations on system performance in terms of cumulative rewards. Our verified results provide provably quantitative certificates for the two questions. We then show that reward martingales can be implemented and trained via neural networks, against different types of control policies. Experimental results demonstrate that our certified bounds tightly enclose simulation outcomes on various DRL-based control systems, indicating the effectiveness and generality of the proposed approach.

cs.AI↗

arXiv2023

Taming Reachability Analysis of DNN-Controlled Systems via Abstraction-Based Training

The intrinsic complexity of deep neural networks (DNNs) makes it challenging to verify not only the networks themselves but also the hosting DNN-controlled systems. Reachability analysis of these systems faces the same challenge. Existing approaches rely on over-approximating DNNs using simpler polynomial models. However, they suffer from low efficiency and large overestimation, and are restricted to specific types of DNNs. This paper presents a novel abstraction-based approach to bypass the crux of over-approximating DNNs in reachability analysis. Specifically, we extend conventional DNNs by inserting an additional abstraction layer, which abstracts a real number to an interval for training. The inserted abstraction layer ensures that the values represented by an interval are indistinguishable to the network for both training and decision-making. Leveraging this, we devise the first black-box reachability analysis approach for DNN-controlled systems, where trained DNNs are only queried as black-box oracles for the actions on abstract states. Our approach is sound, tight, efficient, and agnostic to any DNN type and size. The experimental results on a wide range of benchmarks show that the DNNs trained by using our approach exhibit comparable performance, while the reachability analysis of the corresponding systems becomes more amenable with significant tightness and efficiency improvement over the state-of-the-art white-box approaches.

cs.LG↗

arXiv2019

Proving Expected Sensitivity of Probabilistic Programs with Randomized Variable-Dependent Termination Time

The notion of program sensitivity (aka Lipschitz continuity) specifies that changes in the program input result in proportional changes to the program output. For probabilistic programs the notion is naturally extended to expected sensitivity. A previous approach develops a relational program logic framework for proving expected sensitivity of probabilistic while loops, where the number of iterations is fixed and bounded. In this work, we consider probabilistic while loops where the number of iterations is not fixed, but randomized and depends on the initial input values. We present a sound approach for proving expected sensitivity of such programs. Our sound approach is martingale-based and can be automated through existing martingale-synthesis algorithms. Furthermore, our approach is compositional for sequential composition of while loops under a mild side condition. We demonstrate the effectiveness of our approach on several classical examples from Gambler's Ruin, stochastic hybrid systems and stochastic gradient descent. We also present experimental results showing that our automated approach can handle various probabilistic programs in the literature.

cs.PL↗