Yinan Ni

Yinan Ni


Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
9 February 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
CSChihui Shao
Chihui Shao
YZYun Zi
Yun Zi
YDYingnan Deng
Yingnan Deng
HLHeyao Liu
Heyao Liu
CZChong Zhang
Chong Zhang
YNYinan Ni
Yinan Ni
This paper addresses the problem of text classification models being vulnerable and lacking robustness under adversarial perturbations by proposing a robust text classification method based on large language model calibration. The method builds on a pretrained language model and constructs a multi-stage framework for semantic representation and confidence regulation. It achieves stable optimization of classification results through semantic embedding extraction, calibration adjustment, and consistency constraints. First, the model uses a pretrained encoder to generate context-aware semantic features and applies an attention aggregation mechanism to obtain global semantic representations. Second, a temperature calibration mechanism is introduced to smooth the output probability distribution, reducing the model's sensitivity to local perturbations. Third, adversarial consistency constraints are applied to maintain feature alignment between original and perturbed samples in semantic space, ensuring dynamic preservation of semantic robustness. The method adopts a joint loss function to balance three optimization objectives: classification accuracy, robustness, and confidence. To verify its effectiveness, sensitivity experiments on hyperparameters, environments, and data distributions are conducted. The results show that the model maintains high performance and stability under conditions such as word substitution, noise injection, and class imbalance, significantly outperforming several mainstream baseline models. This study achieves the integration of semantic-level robustness optimization and calibration learning, providing a new approach for building highly reliable text classification systems.

22 December 2025 in Preprints.org
Verified
Preprint
Open access
PDF file
JJJunjie Jiang
Junjie Jiang
CSChihui Shao
Chihui Shao
CZChong Zhang
Chong Zhang
NLNing Lyu
Ning Lyu
YNYinan Ni
Yinan Ni
This study proposes a spatiotemporal framework with cross-node dependency modeling to address anomaly identification in distributed systems under highly dynamic workloads, multi-node coupling, and complex topology changes. The framework first builds a multi-scale temporal representation module that extracts local trends, short-term disturbances, and long-term correlations from monitoring sequences to enhance sensitivity to system dynamics. It then introduces a dynamic structure learning mechanism that generates time-dependent structural graphs based on call chains, resource contention, and link interactions, allowing the model to capture dependency drift caused by topology changes during system operation. On this basis, the framework designs a structure propagation and joint embedding module that fuses temporally enhanced features with structural dependency representations into a unified system state vector, enabling the model to understand anomaly propagation paths and cross-node interactions from a global perspective. A latent-space anomaly measurement function is used to identify possible bottleneck anomalies and improve detection performance for complex system behaviors. To validate the method, extensive experiments are conducted on large-scale open cluster monitoring data, evaluating model performance under hyperparameter sensitivity, environment sensitivity, and data sensitivity settings. The results show that the proposed framework maintains stable performance under structural noise, topology perturbations, and workload fluctuations, and achieves higher accuracy, recall, and F1 scores than multiple representative baselines, demonstrating the advantages of spatiotemporal joint modeling for anomaly detection in distributed systems.

31 December 2025 in Preprints.org
Verified
Preprint
Open access
PDF file
CZChong Zhang
Chong Zhang
CSChihui Shao
Chihui Shao
JJJunjie Jiang
Junjie Jiang
YNYinan Ni
Yinan Ni
XSXiaoxuan Sun
Xiaoxuan Sun
To address the practical challenges of diverse anomaly patterns, strongly coupled dependencies, and high labeling costs in large-scale complex infrastructures, this paper presents an unsupervised anomaly detection method that integrates graph neural networks with Transformer models. The approach learns normal system behavior and identifies deviations without relying on anomaly labels. Infrastructure components are abstracted as nodes in a dependency graph, where nodes are characterized by multiple source observability signals. A graph encoder aggregates neighborhood information to produce structure-enhanced node representations. Self-attention mechanisms are introduced along the temporal dimension to capture long-range dynamic dependencies. This design enables joint modeling of structural relations and temporal evolution. A reconstruction-based training strategy is adopted to constrain the learning of normal patterns. Reconstruction error is used to derive anomaly scores for detection. To ensure reproducibility and ease of deployment, complete specifications of data organization, training procedures, and key hyperparameter settings are provided. Comparative experiments on public benchmarks demonstrate overall advantages across multiple evaluation metrics and confirm the effectiveness of the proposed framework in representing anomaly propagation and temporal drift characteristics in complex systems.