arXiv subjects

Yuanjun Shi

Publications and source records attributed to Yuanjun Shi.

3 recordsLinked to original sources

arXiv2026

SLCA-GRPO: Resolving Cross-Segment Credit Misattribution in Tool-Calling RL

Tool-calling agents produce heterogeneous outputs, interleaving structured tool invocations with user-facing natural language summaries. This output heterogeneity presents a structural failure mode in standard on-policy Reinforcement Learning (RL): algorithms like GRPO indiscriminately broadcast a homogeneous trajectory-level scalar advantage to all tokens. Consequently, gradient noise from summary generation leaks into tool-decision tokens, causing cross-segment credit misattribution and brittle optimization. In this work, we propose SLCA-GRPO, a framework incorporating Segment-Locked Credit Assignment (SLCA). To enable scalable exploration without costly real APIs and stable training, we first construct the Schema-Guided LLM Simulator (SGLS) as foundational training infrastructure. Building on this, SLCA decouples advantage estimation at the structural segment level within a single group of rollouts, without requiring additional rollouts from intermediate states. Supported by Hierarchical Rewards (HierR), SLCA routes execution advantages to tool tokens and preference advantages to summary tokens, eliminating advantage contamination (the dominant cross-segment credit misattribution channel) within each policy update. On a 7B backbone, SLCA-GRPO accelerates convergence and outperforms standard GRPO, ToolPO, and RLTR by +2.53 pp on in-domain evaluation, +1.36 pp on the Berkeley Function-Calling Leaderboard (BFCL), and +9.15 pp on $τ^2$-Bench under the same training budgets, achieving higher accuracy with reduced tool redundancy and costs.

cs.AI↗

arXiv2023

Adaptive End-to-End Metric Learning for Zero-Shot Cross-Domain Slot Filling

Recently slot filling has witnessed great development thanks to deep learning and the availability of large-scale annotated data. However, it poses a critical challenge to handle a novel domain whose samples are never seen during training. The recognition performance might be greatly degraded due to severe domain shifts. Most prior works deal with this problem in a two-pass pipeline manner based on metric learning. In practice, these dominant pipeline models may be limited in computational efficiency and generalization capacity because of non-parallel inference and context-free discrete label embeddings. To this end, we re-examine the typical metric-based methods, and propose a new adaptive end-to-end metric learning scheme for the challenging zero-shot slot filling. Considering simplicity, efficiency and generalizability, we present a cascade-style joint learning framework coupled with context-aware soft label representations and slot-level contrastive representation learning to mitigate the data and label shift problems effectively. Extensive experiments on public benchmarks demonstrate the superiority of the proposed approach over a series of competitive baselines.

cs.CL↗

arXiv2022

Hyperbolic Self-supervised Contrastive Learning Based Network Anomaly Detection

Anomaly detection on the attributed network has recently received increasing attention in many research fields, such as cybernetic anomaly detection and financial fraud detection. With the wide application of deep learning on graph representations, existing approaches choose to apply euclidean graph encoders as their backbone, which may lose important hierarchical information, especially in complex networks. To tackle this problem, we propose an efficient anomaly detection framework using hyperbolic self-supervised contrastive learning. Specifically, we first conduct the data augmentation by performing subgraph sampling. Then we utilize the hierarchical information in hyperbolic space through exponential mapping and logarithmic mapping and obtain the anomaly score by subtracting scores of the positive pairs from the negative pairs via a discriminating process. Finally, extensive experiments on four real-world datasets demonstrate that our approach performs superior over representative baseline approaches.

cs.SI↗