arXiv subjects

Xuanren Chen

Publications and source records attributed to Xuanren Chen.

4 recordsLinked to original sources

arXiv2026

BLPO: Beyond-Local Policy Optimization for Long-Horizon LLM Agents

Long-horizon tasks require large language model (LLM) agents to coordinate interdependent actions, a capability central to automating complex workflows. Stepwise group-based reinforcement learning provides finer credit than trajectory-level supervision, but reliable comparisons must account for differences in interaction history. Enforcing this consistency within the limited rollouts of each task can leave insufficient support for estimating action preferences. We propose Beyond-Local Policy Optimization (BLPO), which estimates task-progress advantages from experience shared across tasks and training updates. We summarize trajectory prefixes into task-progress states and abstract actions by their function, allowing corresponding decisions from different tasks to contribute to shared estimates. To control bias and noise from heterogeneous tasks, we accumulate context-relative returns in a global memory with balanced aggregation and reliability weighting. These global advantages complement local comparisons over normalized actions and trajectory-level credit without a learned critic or additional rollouts. Experiments on ALFWorld and WebShop with 1.5B and 7B policies show improved task success over the evaluated baselines under matched training budgets. Further analyses show faster learning and shorter trajectories with little computational overhead.

cs.AI↗

arXiv2026

FinEvo-Bench: A Longitudinal Benchmark for Self-Evolving Agents in Professional Financial Workflows

Agents used over time encounter recurring professional work: each case requires different evidence and judgment, while the underlying workflow can be reused. Benchmarks built from independent tasks cannot reveal whether an agent turns earlier experience into better procedures for later cases. We introduce FinEvo-Bench, a longitudinal benchmark designed around this structure. It contains 120 open-ended tasks drawn from real cases across 20 business scenes in six financial domains. Each scene contains six substantively different cases that share a professional workflow and an expert-authored rubric for task quality and financial compliance. Constructing and validating the benchmark required approximately 1,200 person-hours. Finance provides a natural test bed because recurring analyses apply shared professional and compliance requirements to heterogeneous inputs, producing case-specific analyses and conclusions. We evaluate four self-evolving agent scaffolds with Qwen3.7-Max on three independently shuffled, globally interleaved task streams. A Claude Code rubric judge backed by Claude Opus~4.6 evaluates all outputs, and paired state-reset controls estimate each scaffold's gain from retained experience. Evolving runs score 9.33--19.37 points higher and trigger 0.12--0.44 fewer compliance issues per task than their paired controls. Paired score gains at within-scene ranks~4--6 exceed those at ranks~1--3 by 6.10--8.70 points. FinEvo-Bench measures whether retained experience improves later professional work under continued use.

cs.AI↗

arXiv2026

DeFA: Dependency-Guided Failure Attribution for LLM Agents

Errors in LLM agent executions and their visible consequences can be separated by many steps, making decisive-error localization a matter of understanding both step content and step dependencies. We introduce DeFA, a dependency-guided framework for agent failure attribution. DeFA first combines protocol relations and semantic dependencies into an event dependency graph spanning the trajectory. It then identifies events that may violate task requirements and traces their sources and subsequent effects to construct a failure propagation graph. Finally, DeFA uses step evidence and the steps' roles in failure propagation to identify the decisive error, responsible agent, and error category. To support long trajectories, DeFA partitions executions into segments and combines the current segment's detailed content with summaries of the other segments, giving local diagnosis access to global execution context. Across Who and When and the Who and When Pro text subset, DeFA achieves the highest responsible-agent and exact step accuracy with all evaluated backbones, and the highest failure-mode accuracy among taxonomy-aligned methods on Pro. Further experiments on image and video trajectories demonstrate its applicability to multimodal failure attribution. Ablations support the contributions of segmentation, the event dependency graph, and the failure propagation graph. Using DeFA's diagnostic feedback for skill evolution in Trace2Skill improves downstream task accuracy by 6-15 percentage points over the native pipeline, showing that the diagnoses can also support agent improvement on subsequent tasks.

cs.AI↗

arXiv2026

VPID: An Integrated Framework for Vulnerability Prioritization and Intrusion Detection in Enterprise Networks

Small enterprises face increasingly serious threats to their internal networks but often lack the financial resources, computing capacity, and specialist staff required to deploy resource intensive security platforms. This paper designs and implements VPID, a lightweight framework for vulnerability prioritization and intrusion detection that consists of two principal modules: controlled vulnerability validation and intelligent intrusion defense. The first module uses OpenVAS for asset mapping and vulnerability identification, applies a decision tree to prioritize vulnerabilities, and employs a rule engine to generate targeted validation payloads. The second module captures network traffic using Scapy, analyzes it through a detection pipeline that combines a decision tree with multinomial Naive Bayes, verifies traffic assessed as high risk using Snort rules, and performs blocking and alerting through iptables. The evaluation uses 550,000 network flow samples containing normal and attack traffic for detector training, together with 15,000 labeled vulnerability records. On the vulnerability ranking test set, the decision tree achieves a precision of 91.8%, a recall of 89.5%, and an F1 score of 90.6%. On an independent test set containing 55,000 traffic samples, the combined detection pipeline achieves a precision of 94.5%, a recall of 88.3%, and an F1 score of 91.3%, while maintaining a false positive rate below 1.5%.

cs.CE↗