arXiv subjects

Zexin Wang

Publications and source records attributed to Zexin Wang.

At least 19 recordsLinked to original sources

arXiv2026

LongRCA Bench: Root-Cause Localization in Long-Horizon Agent Trajectories

In long agent executions, an early error can persist through later actions and checks, while evidence needed to trace its origin is dispersed across the history. Short histories offer limited tests of recovering error origins across substantial subsequent execution. We introduce LongRCA Bench: 1,140 complete failed trajectories from five sources, all human-annotated for responsible roles and earliest decisive root-cause steps. Reference roots precede completion by a median of 48 recorded steps; 28.4% of trajectories contain over 100 subsequent steps. With DeepSeek-V4-Flash on the full benchmark, the strongest of five evaluated baselines achieves 13.2% exact root-step accuracy. We propose Root-Cause Trajectory Attribution (RCTA), a training-free method that organizes original candidate records and explicit handoff instructions for attribution. Segment summaries and a trajectory outline guide candidate retrieval; available handoff records supply upstream instruction context for the final instruction-execution comparison. With the same backbone and scoring protocol, RCTA reaches 24.1% exact root-step accuracy and 51.1% responsible-role accuracy. LongRCA-Mini provides 200 fixed trajectories for lower-cost comparative screening. Even with RCTA, fewer than one quarter of reference roots are recovered exactly.

cs.AI↗

arXiv2026

Agent System Operations: Categorization, Challenges, and Future Directions

As the reasoning capabilities of Large Language Models (LLMs) continue to advance, LLM-based agent systems offer advantages in flexibility and interpretability over traditional systems, garnering increasing attention. However, despite the widespread research interest and industrial application of agent systems, these systems, like their traditional counterparts, frequently encounter anomalies. These anomalies lead to instability and insecurity, hindering their further development. Therefore, a comprehensive and systematic approach to the operation and maintenance of agent systems is urgently needed. Unfortunately, current research on the operations of agent systems is sparse. To address this gap, we have undertaken a survey on agent system operations with the aim of establishing a clear framework for the field, defining the challenges, and facilitating further development. Specifically, this paper begins by systematically defining anomalies within agent systems, categorizing them into intra-agent anomalies and inter-agent anomalies. Next, we introduce a novel and comprehensive operational framework for agent systems, dubbed Agent System Operations (AgentOps). We provide detailed definitions and explanations of its four key stages: monitoring, anomaly detection, root cause localization, and resolution.

cs.MA↗

arXiv2026

Homology of non-matching complexes under edge additions and applications to their Stanley-Reisner ideals

For a bipartite graph $G$ and an integer $t\geq2$, let $\NM_t(G)$ be its $t$-non-matching complex. We prove that adding an edge while preserving bipartiteness induces an injection on reduced homology in degree $2t-3$. Combined with the cyclic-polytope model for non-matching complexes of cycles, this shows that $\NM_t(G)$ has Leray number $2t-2$ whenever $G$ contains a cycle of length at least $2t$. Under the same hypothesis, Hochster's formula yields regularity $2t-1$ for the Stanley-Reisner ideal $I_{\NM_t(G)}$, together with explicit lower bounds for the Betti numbers on its top regularity strand and for its projective dimension. If $G$ contains a $2t$-cycle, we also determine the maximal shifts of $I_{\NM_t(G)}$ through homological degree $|E(G)|-2t+1$. For $G=K_{r,s}$ with $2\leq t\leq r\leq s$, we determine the depth, projective dimension, all maximal shifts, and the unique extremal Betti number of $I_{\NM_t(K_{r,s})}$, thereby settling a conjecture on facet ideals of chessboard complexes.

math.AC↗

arXiv2026

Detect Before You Attribute: Cascade Failure Attribution for Multi-Agent Systems

Large language model (LLM)-based agents have shown strong potential in solving complex tasks through multi-step reasoning, yet they remain vulnerable to execution failures. Accurate failure attribution is therefore critical for improving agent reliability. Existing topology- and spectrum-based methods exploit trajectory structures but often overlook fine-grained semantics, while LLM-based attribution methods capture semantic cues but suffer from long-context degradation over lengthy trajectories. To address these challenges, we propose DUOTRACE, a plug-and-play detection filter for LLM-based failure attribution. DUOTRACE follows a detect-before-attribute paradigm: it first detects anomalous executions and then supplies focused trajectory evidence to downstream LLM-based attribution methods. For effective VAE-based anomaly detection on agent trajectories, DUOTRACE integrates dual-view semantic-structural node representations, a Tree-LSTM-based trajectory encoder, and prefix-chain- and LLM-based data augmentation to handle heterogeneous nodes, hierarchical execution structures, and limited failure data. Experiments with six LLM-based attribution baselines show that DUOTRACE improves agent-level and step-level attribution accuracy by 8.7% and 7.0%, respectively.

cs.AI↗

arXiv2026

Betti numbers of normal edge rings (\bf{I})

We introduce a novel approach named the {\it induced-subgraph approach} for investigating the Betti numbers of normal edge rings. Utilizing this approach, we compute all the multigraded Betti numbers of the edge rings associated with two-ear graphs (Definition~4.1) and compact graphs (Definition~5.1). In particular, we show that for two-ear graphs and compact graphs of type 1 or 2, their multigraded Betti numbers are always equal to the top multigraded Betti numbers of some of their induced subgraphs. In contrast, some of the multigraded Betti numbers of compact graphs of type 3 are not the top multigraded Betti numbers of any of their induced subgraphs. We expect that our approach may apply to many other normal edge rings.

math.AC↗

arXiv2026

Don't Predict, Prioritize: Rethinking GPU Reliability Assessment

The reliability of Graphics Processing Units (GPUs) is a criticalbottleneck for modern large-scale AI infrastructure, where a sin-gle node failure can disrupt synchronous training jobs and causesignificant financial losses. While predictive maintenance is widelyused in other hardware domains, we demonstrate that accuratelypredicting the exact timing of GPU failures is inherently difficult.Through an in-depth analysis of telemetry data from a productioncluster, we find that major GPU failures, including Double Bit Er-rors (DBEs) and GPU Lost events, exhibit strong stochasticity andlow signal-to-noise ratios in time-series telemetry, which makesconventional time-based prediction ineffective. This insight motivates a paradigm shift: instead of attempting topredict the absolute timing of a failure, we propose a more robustapproach focused on ranking nodes by their relative failure risk. Wepropose HeaRank (Health Rank), a Learning-to-Rank (LTR) frame-work that leverages stable historical failure patterns to computea global risk ranking of GPU nodes. Evaluated on a production-scale cluster with thousands of GPUs, HeaRank achieves an AUCof 0.83, significantly outperforming both heuristic baselines andstate-of-the-art ranking algorithms. In online deployment, HeaRanksuccessfully captures 64% of future failures within the top 5% ofranked nodes, compared to only 21% by the incumbent productionsystem. These results suggest that relative risk ranking can serveas a robust alternative in environments where absolute failure pre-diction is inherently limited. Our work highlights the importanceof risk-aware scheduling and proactive resource management inmodern GPU clusters.

cs.DC↗

arXiv2026

A Self-Evolving Agent for Longitudinal Personal Health Management

Personal health management unfolds over repeated encounters, yet most health AI systems treat each request in isolation. We developed HealthClaw, an open-source agent architecture that updates support as a person's routines, preferences, measurements and risks change. It separates shared safety rules and medical knowledge from private longitudinal memory containing profile facts, reusable procedures and episodic traces. After each episode, induction determines what should update the profile, revise a procedure, remain episodic or be excluded. We evaluated HealthClaw with a synthetic year-long benchmark and nine 200-case biomedical tasks. Across 900 longitudinal support probes, answer accuracy increased from 0.2% with current-query prompting to 45.7% with HealthClaw, while prompt-side context exposure was 71.7% lower than with full-history prompting. In 100 privacy probes, HealthClaw produced higher privacy-aware answer quality and fewer unsafe disclosures than both baselines. Across the biomedical tasks, the mean absolute gain in the task-specific primary metric was 27.0 percentage points, and seven gains remained significant after false-discovery-rate correction. These offline benchmarks support governed, self-evolving memory for longitudinal personal health agents, although clinical effectiveness requires prospective evaluation. HealthClaw is publicly available at https://github.com/HC-Guo/HealthClaw.

cs.AI↗

arXiv2026

Mixed Third-Order Flux Laws for Dual Cascade in the Stochastic SQG Equation

We study dual-cascade flux laws for the stochastic forced--dissipative surface quasi-geostrophic (SQG) equation on a large periodic box. For statistically stationary solutions, under a weak anomalous dissipation assumption, we derive rigorous mixed third-order structure-function laws for the dual cascade: a Yaglom-type law for the direct cascade of surface potential energy (SPE) and an antisymmetrized mixed flux law for the inverse cascade of the Hamiltonian. In particular, the inverse Hamiltonian law appears to be new even as an explicit third-order structure-function relation. We also prove Onsager-type obstruction results showing that sufficiently regular stationary families cannot sustain the corresponding non-zero fluxes: $B^s_{3,\infty}$-regularity above the Onsager threshold $1/3$ rules out the direct SPE flux, while sufficient low-frequency Besov regularity rules out the inverse Hamiltonian flux. These results provide a rigorous formulation of the SQG dual-cascade phenomenology in a stochastic stationary setting.

math.AP↗

arXiv2026

UModel: An Agent-Ready Observability Data Modeling Method at Scale

When networked system failures occur, automatically performing Root Cause Analysis (RCA) using observability data is critical for ensuring networked system reliability. Recently, LLM-based agents have shown promise for automating this diagnosis process through advanced reasoning and autonomous exploration. However, existing observability frameworks remain archaic, characterized by fragmented data silos, incompatible schemas, and insufficient semantic metadata, preventing agents from establishing the complex relationships required for effective RCA. To address these challenges, we present UModel, a unified ontological framework that shifts observability from data-centric to object-centric modeling. UModel constructs a virtual ontological layer where heterogeneous telemetry, entities, and expert knowledge are standardized as objects and interconnected via semantic graphs. In addition, we introduce U-SPL, a pipeline-based query interface that enables agents to autonomously explore system topologies and correlate multimodal data. By re-modeling the "AIOps 2025 Challenge" dataset using UModel, the precision of root cause localization improved by 8%, demonstrating that enhanced data organization can significantly increase the accuracy of downstream tasks. UModel provides a scalable modeling framework that, in its deployment at Alibaba Cloud for more than one year, has served tens of thousands of users, sustained millions of operations per second, and delivered sub-second query latency.

cs.SE↗

arXiv2026

KairosVL: Orchestrating Time Series and Semantics for Unified Reasoning

Driven by the increasingly complex and decision-oriented demands of time series analysis, we introduce the Semantic-Conditional Time Series Reasoning task, which extends conventional time series analysis beyond purely numerical modeling to incorporate contextual and semantic understanding. To further enhance the mode's reasoning capabilities on complex time series problems, we propose a two-round reinforcement learning framework: the first round strengthens the mode's perception of fundamental temporal primitives, while the second focuses on semantic-conditioned reasoning. The resulting model, KairosVL, achieves competitive performance across both synthetic and real-world tasks. Extensive experiments and ablation studies demonstrate that our framework not only boosts performance but also preserves intrinsic reasoning ability and significantly improves generalization to unseen scenarios. To summarize, our work highlights the potential of combining semantic reasoning with temporal modeling and provides a practical framework for real-world time series intelligence, which is in urgent demand.

cs.AI↗

arXiv2026

Homological shifts of powers a complementary edge ideal

The homological shift algebra and the projective dimension function of complementary edge ideals are investigated. Let $G$ be a connected graph, and let $I$ be its complementary edge ideal. For bipartite graphs $G$, we show that the projective dimension of $I^s$ increases strictly with $s$ until reaching its maximum value. For trees and cycles, explicit expressions for the projective dimension of $I^s$ are provided, along with detailed descriptions of their homological shift algebras. In particular, it is shown that the $i$-th homological shift algebra of such ideals is generated in degree at most $i$. Additionally, we prove that if $G$ is a tree, then the homological shift ideal $\mathrm{HS}_i(I^i)$, when divided by a suitable monomial, becomes a Veronese-type ideal, and every Veronese-type ideal arises in this manner.

math.AC↗

arXiv2025

DistilCLIP-EEG: Enhancing Epileptic Seizure Detection Through Multi-modal Learning and Knowledge Distillation

Epilepsy is a prevalent neurological disorder marked by sudden, brief episodes of excessive neuronal activity caused by abnormal electrical discharges, which may lead to some mental disorders. Most existing deep learning methods for epilepsy detection rely solely on unimodal EEG signals, neglecting the potential benefits of multimodal information. To address this, we propose a novel multimodal model, DistilCLIP-EEG, based on the CLIP framework, which integrates both EEG signals and text descriptions to capture comprehensive features of epileptic seizures. The model involves an EEG encoder based on the Conformer architecture as a text encoder, the proposed Learnable BERT (BERT-LP) as prompt learning within the encoders. Both operate in a shared latent space for effective cross-modal representation learning. To enhance efficiency and adaptability, we introduce a knowledge distillation method where the trained DistilCLIP-EEG serves as a teacher to guide a more compact student model to reduce training complexity and time. On the TUSZ, AUBMC, and CHB-MIT datasets, both the teacher and student models achieved accuracy rates exceeding 97%. Across all datasets, the F1-scores were consistently above 0.94, demonstrating the robustness and reliability of the proposed framework. Moreover, the student model's parameter count and model size are approximately 58.1% of those of the teacher model, significantly reducing model complexity and storage requirements while maintaining high performance. These results highlight the potential of our proposed model for EEG-based epilepsy detection and establish a solid foundation for deploying lightweight models in resource-constrained settings.

cs.LG↗

arXiv2025

ViTs: Teaching Machines to See Time Series Anomalies Like Human Experts

Web service administrators must ensure the stability of multiple systems by promptly detecting anomalies in Key Performance Indicators (KPIs). Achieving the goal of "train once, infer across scenarios" remains a fundamental challenge for time series anomaly detection models. Beyond improving zero-shot generalization, such models must also flexibly handle sequences of varying lengths during inference, ranging from one hour to one week, without retraining. Conventional approaches rely on sliding-window encoding and self-supervised learning, which restrict inference to fixed-length inputs. Large Language Models (LLMs) have demonstrated remarkable zero-shot capabilities across general domains. However, when applied to time series data, they face inherent limitations due to context length. To address this issue, we propose ViTs, a Vision-Language Model (VLM)-based framework that converts time series curves into visual representations. By rescaling time series images, temporal dependencies are preserved while maintaining a consistent input size, thereby enabling efficient processing of arbitrarily long sequences without context constraints. Training VLMs for this purpose introduces unique challenges, primarily due to the scarcity of aligned time series image-text data. To overcome this, we employ an evolutionary algorithm to automatically generate thousands of high-quality image-text pairs and design a three-stage training pipeline consisting of: (1) time series knowledge injection, (2) anomaly detection enhancement, and (3) anomaly reasoning refinement. Extensive experiments demonstrate that ViTs substantially enhance the ability of VLMs to understand and detect anomalies in time series data. All datasets and code will be publicly released at: https://anonymous.4open.science/r/ViTs-C484/.

cs.LG↗

arXiv2025

A Survey on AgentOps: Categorization, Challenges, and Future Directions

cs.AI↗

arXiv2025

EffiVLM-BENCH: A Comprehensive Benchmark for Evaluating Training-Free Acceleration in Large Vision-Language Models

Large Vision-Language Models (LVLMs) have achieved remarkable success, yet their significant computational demands hinder practical deployment. While efforts to improve LVLM efficiency are growing, existing methods lack comprehensive evaluation across diverse backbones, benchmarks, and metrics. In this work, we systematically evaluate mainstream acceleration techniques for LVLMs, categorized into token and parameter compression. We introduce EffiVLM-Bench, a unified framework for assessing not only absolute performance but also generalization and loyalty, while exploring Pareto-optimal trade-offs. Our extensive experiments and in-depth analyses offer insights into optimal strategies for accelerating LVLMs. We open-source code and recipes for EffiVLM-Bench to foster future research.

cs.CL↗

arXiv2025

Betti Numbers of Edge Ideals of Weighted Oriented Crown Graphs

We compute the multigraded Betti numbers of edge ideals for weighted oriented crown graphs using a novel approach. This approach, which we still call the \emph{induced subgraph approach}, originates from our prior work on computing Betti numbers of normal edge rings(see \cite{WL}). Notably, we prove that the total Betti numbers of edge ideals for weighted oriented crown graphs are independent of the weight function.

math.AC↗

arXiv2025

Betti numbers of normal edge rings (II)

We compute the Betti numbers of the edge rings of multi-path graphs using the \emph{induced-subgraph approach} introduced in \cite{WL1}. Here, a multi-path graph refers to a simple graph composed of several paths that have the same starting point and the same ending point. Special cases include the graph $G_{r,d}$ introduced in \cite{GHK}, the graph $G_{r,s,d}$ introduced in \cite{NN}, and the graph $B_{\underline{\ell},h}$ introduced in \cite{LZ}. In particular, we show that all the multi-graded Betti numbers of a multi-path graph are the top multi-graded Betti numbers of some of its induced subgraphs.

math.AC↗

arXiv2025

The Multiplicity of Powers of a Class of Non-Square-free Monomial Ideals

Let $R = \mathbb{K}[x_1, \ldots, x_n]$ be a polynomial ring over a field $\mathbb{K}$, and let $I \subseteq R$ be a monomial ideal of height $h$. We provide a formula for the multiplicity of the powers of $I$ when all the primary ideals of height $h$ in the irredundant reduced primary decomposition of $I$ are irreducible. This is a generalization of \cite[Theorem 1.1]{TV}. Furthermore, we present a formula for the multiplicity of powers of special powers of monomial ideals that satisfy the aforementioned conditions. Here, for an integer $m>0$, the $m$-th special power of a monomial ideal refers to the ideal generated by the $m$-th powers of all its minimal generators. Finally, we explicitly provide a formula for the multiplicity of powers of special powers of edge ideals of weighted oriented graphs.

math.AC↗