arXiv subjects

Yuanchi Ma

Publications and source records attributed to Yuanchi Ma.

6 recordsLinked to original sources

arXiv2026

Is External Database Protection Static in Retrieval-Augmented Generation? Rethinking Privacy Preservation under Dynamic Queries

Retrieval-augmented generation (RAG) enhances large language models via external document retrieval, but retrieved contexts may leak sensitive information. Current privacy protection methods typically rely on a document-level static risk assumption, treating all retrieved documents as having the same privacy leakage risk. However, this assumption overlooks a fundamental characteristic of RAG: the privacy risk of a document is highly dependent on the user's query, making privacy leakage inherently query-driven and dynamic. To address this challenge, we propose a Prompt-Aware Dynamic Hierarchical Differential Privacy framework (PA-HDP) for privacy-preserving RAG. PA-HDP first performs a prompt-aware risk hierarchy to dynamically assess privacy risks under different queries. It then applies adaptive sensitive entity replacement and exponential mechanism-based text selection to provide differentiated privacy protection while preserving semantic utility. By protecting only the content that is truly sensitive under a given query, PA-HDP minimizes unnecessary modifications to the retrieval corpus. Extensive experiments on benchmark datasets demonstrate that PA-HDP significantly reduces privacy leakage while maintaining high retrieval quality, achieving a better privacy-utility trade-off than prior methods.

cs.CR↗

arXiv2026

Creative Convergence or Imitation? Genre-Specific Homogeneity in LLM-Generated Chinese Literature

Large Language Models (LLMs) have demonstrated remarkable capabilities in narrative generation. However, they often produce structurally homogenized stories, frequently following repetitive arrangements and combinations of plot events along with stereotypical resolutions. In this paper, we propose a novel theoretical framework for analysis by incorporating Proppian narratology and narrative functions. This framework is used to analyze the composition of narrative texts generated by LLMs to uncover their underlying narrative logic. Taking Chinese web literature as our research focus, we extend Propp's narrative theory, defining 34 narrative functions suited to modern web narrative structures. We further construct a human-annotated corpus to support the analysis of narrative structures within LLM-generated text. Experiments reveal that the primary reasons for the singular narrative logic and severe homogenization in generated texts are that current LLMs are unable to correctly comprehend the meanings of narrative functions and instead adhere to rigid narrative generation paradigms.

cs.CL↗

arXiv2025

Why Federated Optimization Fails to Achieve Perfect Fitting? A Theoretical Perspective on Client-Side Optima

Federated optimization is a constrained form of distributed optimization that enables training a global model without directly sharing client data. Although existing algorithms can guarantee convergence in theory and often achieve stable training in practice, the reasons behind performance degradation under data heterogeneity remain unclear. To address this gap, the main contribution of this paper is to provide a theoretical perspective that explains why such degradation occurs. We introduce the assumption that heterogeneous client data lead to distinct local optima, and show that this assumption implies two key consequences: 1) the distance among clients' local optima raises the lower bound of the global objective, making perfect fitting of all client data impossible; and 2) in the final training stage, the global model oscillates within a region instead of converging to a single optimum, limiting its ability to fully fit the data. These results provide a principled explanation for performance degradation in non-iid settings, which we further validate through experiments across multiple tasks and neural network architectures. The framework used in this paper is open-sourced at: https://github.com/NPCLEI/fedtorch.

cs.LG↗

arXiv2025

SEMPO: Lightweight Foundation Models for Time Series Forecasting

The recent boom of large pre-trained models witnesses remarkable success in developing foundation models (FMs) for time series forecasting. Despite impressive performance across diverse downstream forecasting tasks, existing time series FMs possess massive network architectures and require substantial pre-training on large-scale datasets, which significantly hinders their deployment in resource-constrained environments. In response to this growing tension between versatility and affordability, we propose SEMPO, a novel lightweight foundation model that requires pretraining on relatively small-scale data, yet exhibits strong general time series forecasting. Concretely, SEMPO comprises two key modules: 1) energy-aware SpEctral decomposition module, that substantially improves the utilization of pre-training data by modeling not only the high-energy frequency signals but also the low-energy yet informative frequency signals that are ignored in current methods; and 2) Mixture-of-PrOmpts enabled Transformer, that learns heterogeneous temporal patterns through small dataset-specific prompts and adaptively routes time series tokens to prompt-based experts for parameter-efficient model adaptation across different datasets and domains. Equipped with these modules, SEMPO significantly reduces both pre-training data scale and model size, while achieving strong generalization. Extensive experiments on two large-scale benchmarks covering 16 datasets demonstrate the superior performance of SEMPO in both zero-shot and few-shot forecasting scenarios compared with state-of-the-art methods. Code and data are available at https://github.com/mala-lab/SEMPO.

cs.LG↗

arXiv2024

Long text outline generation: Chinese text outline based on unsupervised framework and large language mode

Outline generation aims to reveal the internal structure of a document by identifying underlying chapter relationships and generating corresponding chapter summaries. Although existing deep learning methods and large models perform well on small- and medium-sized texts, they struggle to produce readable outlines for very long texts (such as fictional works), often failing to segment chapters coherently. In this paper, we propose a novel outline generation method for Chinese, combining an unsupervised framework with large models. Specifically, the method first generates chapter feature graph data based on entity and syntactic dependency relationships. Then, a representation module based on graph attention layers learns deep embeddings of the chapter graph data. Using these chapter embeddings, we design an operator based on Markov chain principles to segment plot boundaries. Finally, we employ a large model to generate summaries of each plot segment and produce the overall outline. We evaluate our model based on segmentation accuracy and outline readability, and our performance outperforms several deep learning models and large models in comparative evaluations.

cs.AI↗

arXiv2024

Masked AutoEncoder for Graph Clustering without Pre-defined Cluster Number k

Graph clustering algorithms with autoencoder structures have recently gained popularity due to their efficient performance and low training cost. However, for existing graph autoencoder clustering algorithms based on GCN or GAT, not only do they lack good generalization ability, but also the number of clusters clustered by such autoencoder models is difficult to determine automatically. To solve this problem, we propose a new framework called Graph Clustering with Masked Autoencoders (GCMA). It employs our designed fusion autoencoder based on the graph masking method for the fusion coding of graph. It introduces our improved density-based clustering algorithm as a second decoder while decoding with multi-target reconstruction. By decoding the mask embedding, our model can capture more generalized and comprehensive knowledge. The number of clusters and clustering results can be output end-to-end while improving the generalization ability. As a nonparametric class method, extensive experiments demonstrate the superiority of \textit{GCMA} over state-of-the-art baselines.

cs.LG↗