Selected work

Representative Papers

Statistical Downscaling via High-Dimensional Distribution Matching with Generative Models

Dec 11, 2024arXiv.org

High-resolution climate information is critically lacking for kilometer-scale regional climate risk assessment; existing statistical downscaling methods suffer from poor scalability, narrow hazard adaptability, and inability to model physical dependencies among multivariate climate fields. Method: We propose GenBCSR, a two-stage generative framework that—uniquely—decouples statistical downscaling into bias correction and statistical super-resolution, formulated as two high-dimensional distribution alignment tasks, eliminating reliance on pixel-wise paired labels. Leveraging diffusion- or flow-based probabilistic distribution matching, it enables efficient, physics-consistent upscaling under unsupervised or weakly supervised settings. Results: GenBCSR reduces prediction error by 4–5× for compound extreme-event indices (e.g., 99th-percentile metrics) versus conventional methods, achieving superior accuracy, computational efficiency, and physical interpretability.

2 citationsRead paper

Vision-Force Admittance Learning for Peg Insertion into a Movable Hole

Sep 12, 2026

为解决动态环境中精确操作问题，提出视觉-力顺应性学习框架VFAL，融合视觉反馈与力模型，实现高精度适应性插入。

0 citationsRead paper

Learning Agent-based Model Predictive Control for Holistic Vehicle Performance

Sep 10, 2026

研究提出学习代理模型预测控制(LAMPC)，结合基于模型的AMPC和数据驱动方法，通过高斯过程回归预测未知贡献，提高多代理系统整体车辆性能。

0 citationsRead paper

Context-Aware Intelligent Vehicles

Aug 31, 2026

论文探讨了通过将情境因素作为下一代车辆系统的核心原则来解决智能车辆在复杂环境中运行时面临的准确性、延迟、成本和可靠性挑战。

0 citationsRead paper

Budgeted Act-or-Defer Multi-Agent LLM Deliberation with Local Reliability Bounds

Jun 28, 2026

This work addresses the challenge of dynamically deciding whether a multi-agent large language model system should execute an answer or defer to human review under a limited error budget. The problem is formalized as a budget-constrained “act-or-defer” decision framework, where debate prefixes are mapped to low-dimensional states, and a k-nearest-neighbor lower confidence bound on state-conditional correctness is computed using calibration data. An action is executed only when this bound exceeds a user-specified threshold. The authors introduce a novel conditionally safe assurance framework that decomposes the total error budget into three interpretable components: calibration failure, residual risk, and representation gap, enabling falsifiable diagnostics and task-difficulty-normalized budget allocation. Evaluated across six benchmarks, the method achieves 84% automation rate and 96% execution accuracy using only 9–12% of the error budget, substantially outperforming nine baselines.

0 citationsRead paper

Recent publications

Latest Papers

Vision-Force Admittance Learning for Peg Insertion into a Movable Hole

Sep 12, 2026

为解决动态环境中精确操作问题，提出视觉-力顺应性学习框架VFAL，融合视觉反馈与力模型，实现高精度适应性插入。

0 citationsRead paper

Learning Agent-based Model Predictive Control for Holistic Vehicle Performance

Sep 10, 2026

研究提出学习代理模型预测控制(LAMPC)，结合基于模型的AMPC和数据驱动方法，通过高斯过程回归预测未知贡献，提高多代理系统整体车辆性能。

0 citationsRead paper

Context-Aware Intelligent Vehicles

Aug 31, 2026

论文探讨了通过将情境因素作为下一代车辆系统的核心原则来解决智能车辆在复杂环境中运行时面临的准确性、延迟、成本和可靠性挑战。

0 citationsRead paper

Budgeted Act-or-Defer Multi-Agent LLM Deliberation with Local Reliability Bounds

Jun 28, 2026

This work addresses the challenge of dynamically deciding whether a multi-agent large language model system should execute an answer or defer to human review under a limited error budget. The problem is formalized as a budget-constrained “act-or-defer” decision framework, where debate prefixes are mapped to low-dimensional states, and a k-nearest-neighbor lower confidence bound on state-conditional correctness is computed using calibration data. An action is executed only when this bound exceeds a user-specified threshold. The authors introduce a novel conditionally safe assurance framework that decomposes the total error budget into three interpretable components: calibration failure, residual risk, and representation gap, enabling falsifiable diagnostics and task-difficulty-normalized budget allocation. Evaluated across six benchmarks, the method achieves 84% automation rate and 96% execution accuracy using only 9–12% of the error budget, substantially outperforming nine baselines.

0 citationsRead paper

A Differentiable GPU-Accelerated Finite Element Framework for Inverse Characterization of Finite-Strain Anisotropic Plasticity

Jun 15, 2026

This study addresses the challenge of efficiently and accurately inverting finite-strain anisotropic elastoplastic constitutive models in high-dimensional parameter spaces. To this end, we propose a JAX-based differentiable, GPU-accelerated finite element framework that tightly integrates automatic differentiation with finite element computations, eliminating the need for manual gradient derivation and enabling PDE-constrained inverse parameter identification. By leveraging heterogeneous specimen geometries inspired by topology optimization and full-field displacement measurements, the approach substantially reduces experimental dependency. Implemented on a single H100 GPU, the framework achieves up to 9.4× speedup over an Abaqus implementation running on a 24-core CPU and successfully recovers both homogeneous and spatially varying anisotropic yield and hardening parameters, demonstrating its efficiency and feasibility for high-dimensional inverse problems.

0 citationsRead paper