arXiv subjects

Lili Xie

Publications and source records attributed to Lili Xie.

3 recordsLinked to original sources

arXiv2026

Scenario-conditioned flow matching for probabilistic generation of three-component ground-motion waveforms

Performance-based seismic risk assessment requires three-component acceleration histories compatible with specified source, path, and site conditions. Conventional ground-motion prediction equations provide scalar intensity measures, while many generative waveform models learn amplitude and waveform shape within a single high-dimensional target. We present WaveFlowGMM, a two-stage probabilistic ground-motion model that uses peak ground acceleration (PGA) as an amplitude interface between scenario conditioning and waveform generation. The amplitude stage uses physics-informed symbolic learning to estimate component-wise PGA medians and a full cross-component covariance. The waveform stage uses few-step AlphaFlow in an invertible wavelet-packet coefficient space to generate normalised three-component histories that are rescaled by sampled PGA. Tests on an event-level NGA-West2 holdout set show that the generated motions recover the main magnitude, distance, and site scaling, keep peak and spectral residuals close to zero, preserve three-component amplitude dependence, and yield velocity and displacement histories without systematic drift after integration of the generated three-component acceleration histories. The framework provides an interpretable and computationally efficient candidate component for waveform-level seismic hazard and risk analysis.

physics.geo-ph↗

arXiv2025

Beyond Static LLM Policies: Imitation-Enhanced Reinforcement Learning for Recommendation

Recommender systems (RecSys) have become critical tools for enhancing user engagement by delivering personalized content across diverse digital platforms. Recent advancements in large language models (LLMs) demonstrate significant potential for improving RecSys, primarily due to their exceptional generalization capabilities and sophisticated contextual understanding, which facilitate the generation of flexible and interpretable recommendations. However, the direct deployment of LLMs as primary recommendation policies presents notable challenges, including persistent latency issues stemming from frequent API calls and inherent model limitations such as hallucinations and biases. To address these issues, this paper proposes a novel offline reinforcement learning (RL) framework that leverages imitation learning from LLM-generated trajectories. Specifically, inverse reinforcement learning is employed to extract robust reward models from LLM demonstrations. This approach negates the need for LLM fine-tuning, thereby substantially reducing computational overhead. Simultaneously, the RL policy is guided by the cumulative rewards derived from these demonstrations, effectively transferring the semantic insights captured by the LLM. Comprehensive experiments conducted on two benchmark datasets validate the effectiveness of the proposed method, demonstrating superior performance when compared against state-of-the-art RL-based and in-context learning baselines. The code can be found at https://github.com/ArronDZhang/IL-Rec.

cs.IR↗

arXiv2025

MARCO: A Cooperative Knowledge Transfer Framework for Personalized Cross-domain Recommendations

Recommender systems frequently encounter data sparsity issues, particularly when addressing cold-start scenarios involving new users or items. Multi-source cross-domain recommendation (CDR) addresses these challenges by transferring valuable knowledge from multiple source domains to enhance recommendations in a target domain. However, existing reinforcement learning (RL)-based CDR methods typically rely on a single-agent framework, leading to negative transfer issues caused by inconsistent domain contributions and inherent distributional discrepancies among source domains. To overcome these limitations, MARCO, a Multi-Agent Reinforcement Learning-based Cross-Domain recommendation framework, is proposed. It leverages cooperative multi-agent reinforcement learning, where each agent is dedicated to estimating the contribution from an individual source domain, effectively managing credit assignment and mitigating negative transfer. In addition, an entropy-based action diversity penalty is introduced to enhance policy expressiveness and stabilize training by encouraging diverse agents' joint actions. Extensive experiments across four benchmark datasets demonstrate MARCO's superior performance over state-of-the-art methods, highlighting its robustness and strong generalization capabilities. The code is at https://github.com/xiewilliams/MARCO.

cs.IR↗