Yuanzhang Xiao

Prof. Yuanzhang Xiao


Research Focus & Metrics

Academic Career Length

14Years (2013-Present)

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

Representative Publications

View All Publications
10 July 2025 in ACM Transactions on Information Systems
Verified
Journal Article
Open access
SLSichun Luo
Sichun Luo
BHBowei He
Bowei He
HZHaohan Zhao
Haohan Zhao
WSWei Shao
Wei Shao
YQYanlin Qi
Yanlin Qi
LSLinqi Song
Linqi Song
Large language models (LLMs) have demonstrated remarkable capabilities and have been extensively deployed across various domains, including recommender systems. Prior research has employed specialized prompts to leverage the in-context learning capabilities of LLMs for recommendation purposes. More recent studies have utilized instruction tuning techniques to align LLMs with human preferences, promising more effective recommendations. However, existing methods suffer from several limitations. The full potential of LLMs is not fully elicited due to low-quality tuning data and the overlooked integration of conventional recommender signals. Furthermore, LLMs may generate inconsistent responses for different ranking tasks in the recommendation, potentially leading to unreliable results. In this article, we introduce Ranker for top- k Recommendations (RecRanker), tailored for instruction tuning LLMs to serve as the Ranker for top- k Recommendations. Specifically, we introduce importance-aware sampling, clustering-based sampling, and penalty for repetitive sampling for sampling high-quality, representative, and diverse training data. To enhance the prompt, we introduce a position shifting strategy to mitigate position bias and augment the prompt with auxiliary information from conventional recommendation models, thereby enriching the contextual understanding of the LLM. Subsequently, we utilize the sampled data to assemble an instruction-tuning dataset with the augmented prompts comprising three distinct ranking tasks: pointwise, pairwise, and listwise rankings. We further propose a hybrid ranking method to enhance the model performance by ensembling these ranking tasks. Our empirical evaluations demonstrate the effectiveness of our proposed RecRanker in both direct and sequential recommendation scenarios. 1

31 October 2024 in ACM Transactions on Intelligent Systems and Technology
Verified
Journal Article
Open access
SLSichun Luo
Sichun Luo
YXYuanzhang Xiao
Yuanzhang Xiao
XZXinyi Zhang
Xinyi Zhang
YLYang Liu
Yang Liu
Wenbo DingWenbo Ding
Wenbo Ding
LSLinqi Song
Linqi Song
Federated recommendation systems employ federated learning techniques to safeguard user privacy by transmitting model parameters instead of raw user data between user devices and the central server. Nevertheless, the current federated recommender system faces three significant challenges: (1) data heterogeneity: the heterogeneity of users’ attributes and local data necessitates the acquisition of personalized models to improve the performance of federated recommendation; (2) model performance degradation: the privacy-preserving protocol design in the federated recommendation, such as pseudo item labeling and differential privacy, would deteriorate the model performance; (3) communication bottleneck: the standard federated recommendation algorithm can have a high communication overhead. Previous studies have attempted to address these issues, but none have been able to solve them simultaneously. In this article, we propose a novel framework, named PerFedRec++ , to enhance the personalized federated recommendation with self-supervised pre-training. Specifically, we utilize the privacy-preserving mechanism of federated recommender systems to generate two augmented graph views, which are used as contrastive tasks in self-supervised graph learning to pre-train the model. Pre-training enhances the performance of federated models by improving the uniformity of representation learning. Also, by providing a better initial state for federated training, pre-training makes the overall training converge faster, thus alleviating the heavy communication burden. We then construct a collaborative graph to learn the client representation through a federated graph neural network. Based on these learned representations, we cluster users into different user groups and learn personalized models for each cluster. Each user learns a personalized model by combining the global federated model, the cluster-level federated model, and its own fine-tuned local model. Experiments on three real-world datasets show that our proposed method achieves superior performance over existing methods.

15 September 2020 in IEEE Transactions on Smart Grid
Verified
Journal Article
Closed access
DEDouglas Ellman
Douglas Ellman
YXYuanzhang Xiao
Yuanzhang Xiao
We study baseline-based demand response (DR) programs. In such programs, customers get rebates based on how much they reduce electricity consumption during DR events relative to a “baseline”, where this baseline is determined by their consumption during previous non-event days. Customers, or automated controls working on their behalf, can achieve higher DR payments by decreasing consumption during DR events (desired behavior), and by increasing consumption during non-event times (baseline manipulation). Importantly, the customers have imperfect knowledge of when future demand response events will occur. To understand customers’ incentives for baseline manipulation, we present a novel multi-stage stochastic dynamic programming model that optimizes customer actions for maximum expected rewards under uncertain event schedules. Analytical results for special cases show fundamental drivers of customer incentives. Simulation results reveal incentives to manipulate baselines and impacts to program performance for a realistic baseline-based demand response program, and how program and customer parameters affect incentives.

16 January 2026 in IEEE Journal of Selected Topics in Signal Processing
Verified
Journal Article
Closed access
SLSichun Luo
Sichun Luo
YYYuxuan Yao
Yuxuan Yao
BHBowei He
Bowei He
WSWei Shao
Wei Shao
JXJian Xu
Jian Xu
LSLinqi Song
Linqi Song
Conventional recommender systems and Large Language Model (LLM)-based recommender systems each have their strengths and weaknesses. While conventional recommendation methods excel at mining collaborative information and modeling sequential behavior, they struggle with data sparsity and the long-tail problem. LLM, on the other hand, is proficient at utilizing rich textual contexts but faces challenges in mining collaborative or sequential information. Despite their individual successes, there is a significant gap in leveraging their ensemble potential to enhance recommendation performance. In this paper, we introduce a general and model-agnostic framework known as Large language models with mutual augmentation and adaptive aggregation for Recommendation (Llama4Rec), aiming to bridge this gap via explicitly ensemble LLM and conventional recommendation model for more effective recommendation. We propose data augmentation and prompt augmentation strategies tailored to enhance the conventional recommendation model and LLM respectively. An adaptive aggregation module is adopted to combine the predictions of both kinds of models to refine the final recommendation results. Empirical studies on three datasets validate the superiority of Llama4Rec, demonstrating significant improvements in recommendation performance.

21 October 2023
Verified
Conference Paper
Open access
SLSichun Luo
Sichun Luo
CMChen Ma
Chen Ma
YXYuanzhang Xiao
Yuanzhang Xiao
LSLinqi Song
Linqi Song
The ubiquitous long-tail distribution of inherent user behaviors results in worse recommendation performance for the items with fewer user records (i.e., tail items) than those with richer ones (i.e., head items). Graph-based recommendation methods (e.g., using graph neural networks) have recently emerged as a powerful tool for recommender systems, often outperforming traditional methods. However, existing techniques for alleviating the long-tail problem mainly focus on traditional methods. There is a lack of graph-based methods that can efficiently deal with the long-tail problem. In this paper, we propose a novel approach, Graph Augmentation for Long-tail Recommendation (GALORE), which can be plugged into any graph-based recommendation models to improve the performance for tail items. GALORE incorporates an edge addition module that enriches the graph's connectivity for tail items by injecting additional item-to-item edges. To further balance the graph structure, GALORE utilizes a degree-aware edge dropping strategy, preserving the more valuable edges from the tail items while selectively discarding less informative edges from the head items. Beyond structural augmentation, we synthesize new data samples, thereby addressing the data scarcity issue for tail items. We further introduce a two-stage training strategy to facilitate the learning for both head and tail items. Comprehensive empirical studies conducted on four datasets show that GALORE outperforms existing methods in terms of the performance for tail items as well as the overall performance.

19 April 2021 in IEEE Transactions on Green Communications and Networking
Verified
Journal Article
Closed access
TLTan Li
Tan Li
YXYuanzhang Xiao
Yuanzhang Xiao
LSLinqi Song
Linqi Song
Residential demand side management (DSM) is a promising technique in smart grids to improve the power system robustness and to reduce the energy cost. However, the ongoing paradigm shift of computation, such as mobile edge computing for smart home, poses a big challenge to residential DSM. Therefore, it is important to schedule the new smart home computing tasks and traditional DSM in a smart way. In this paper, we investigate an integrated home energy management system (HEMS) that participates in a DSM program and implements smart home computation tasks by offloading tasks with the help of a Smart Home Operation Platform (SHOP). The goal of HEMS is to maximize the user's expected total reward, defined as the reward from completing computing tasks minus the cost of energy consumption, execution delay, running the SHOP servers, and the penalty of violating the DSM requirements. We solve this task scheduling based DSM problem using a deep reinforcement learning method. The DSM program considered in this paper requires the household to reduce a certain amount of energy consumption within a specified time window, which, in stark contrast to the well-studied real-time pricing, results in a long-term temporal interdependence and thus a high-dimensional state space in our formulated problem. To address this challenge, we use the Deep Deterministic Policy Gradient (DDPG) method to characterize the high-dimensional state space and action space, which uses deep neural networks to estimate the state and to generate the action. Experimental results show that our proposed method achieves better performance gains over reasonable baselines.