Institution profile

Clemson University

Academic institutionnorthamerica · us
Official website
Research library373linked papers
Opportunities0open roles
Selected work

Representative Papers

OpenDecoder: Open Large Language Model Decoding to Incorporate Document Quality in RAG

Jan 13, 2026

This work proposes OpenDecoder, a novel framework that addresses the challenge of inconsistent retrieval quality in Retrieval-Augmented Generation (RAG) systems, which often undermines answer accuracy. OpenDecoder is the first approach to explicitly integrate multi-dimensional document quality signals—including relevance scores, ranking positions, and query performance prediction metrics—directly into the decoding process of large language models (LLMs), enabling quality-aware generation control. The framework is highly flexible, allowing seamless incorporation of arbitrary external quality indicators and compatibility with various LLM post-training objectives. Extensive experiments across five benchmark datasets demonstrate that OpenDecoder significantly outperforms existing baselines, substantially enhancing the robustness of RAG systems against noisy contexts and improving the reliability of generated responses.

2 citationsRead paper

Numerical Simulation Informed Rapid Cure Process Optimization of Composite Structures using Constrained Bayesian Optimization

May 30, 2025

To address structural deformation in composite material curing arising from thermo-chemo-mechanical coupling, this paper proposes an efficient process optimization method integrating Gaussian process surrogate modeling with constrained Bayesian optimization (cBO). The objective is to minimize curing-induced deformation while strictly satisfying critical process constraints—particularly complete cure. This work represents the first systematic application of cBO to such multi-physics coupled optimization problems. Compared with conventional genetic algorithms requiring over 1,000 iterations, cBO achieves convergence in fewer than 50 iterations for both flat-plate and L-shaped composite structures, yielding over 20× computational speedup (>96% reduction in function evaluations) and maintaining a prediction error below 4%. The approach successfully attains dual optimization goals: strict constraint satisfaction and global minimization of residual deformation.

1 citationsRead paper

Patient flow networks absorb healthcare stress during pandemic crises

Oct 08, 2024arXiv.org

Large-scale disasters (e.g., pandemics) severely undermine healthcare system resilience, particularly through regional imbalances in bed capacity, workforce, and medical supplies. Existing studies predominantly focus on aggregate system-level outcomes, overlooking the critical buffering role of inter-regional coordination. Method: Leveraging billions of electronic health records, we construct a nationwide patient mobility network and propose—firstly—the metric “healthcare system absorptivity”: the network’s capacity to dynamically redistribute patients and alleviate localized strain. We integrate complex network modeling with a dynamic flow assessment framework under crisis conditions. Contribution/Results: During the COVID-19 pandemic, inter-regional patient mobility increased by 3.89%, yielding an average absorptivity of 0.21—10% higher than pre-pandemic levels. Networks exhibiting high connectivity and structural heterogeneity further enhanced resilience by up to 10%. These findings empirically validate optimized inter-regional patient flow as an effective strategy for strengthening healthcare system resilience.

1 citationsRead paper
Recent publications

Latest Papers