Geometric Signatures of Conceptual Reorganization: A Counterfactual Embedding Framework for Detecting Scientific Revolutions
本文通过引入文档嵌入几何和反事实消融框架，提供了一种量化检测科学革命的方法，用以测量单个概念对科学知识组织的影响。
0 citationsRead paper
Selected work
Representative Papers
Latent unified smooth Hamiltonians for excited state chemistry
本文提出一种神经网络架构，通过学习电子态哈密顿量的隐式基表示来统一处理分子系统的基态和激发态问题。
0 citationsRead paper
Panda Diplomacy: Foundation Model Pre-training across Particle Imaging Detectors for High Energy and Nuclear Physics
本文通过点云自蒸馏框架解决了粒子和核物理中基础模型跨传感器复用的问题，提高了模型在不同探测器上的性能。
0 citationsRead paper
Unknown Unknowns: Model Misspecification in Machine Learning for Physics
This study addresses the risks of unknown unknowns arising from model misspecification in physical inverse problems by proposing an iterative diagnosis and mitigation framework. Treating misspecification as an opportunity for discovery, this work establishes a closed-loop detection-mitigation analytical paradigm that integrates complementary diagnostics, iterative updating, and robustness analysis strategies. Consequently, this research develops a systematic methodology for managing unknown unknowns, effectively enhancing model robustness against unforeseen biases and significantly improving the reliability of physical measurements. Ultimately, the proposed framework provides a novel safety assurance mechanism for solving complex inverse problems, ensuring greater confidence in computational reconstructions where model fidelity cannot be fully guaranteed a priori.
0 citationsRead paper
Contrastive Learning for Interpretable Anomaly Detection at Collider Experiments
This study addresses the challenges of score interpretability and strong energy dependence in collider anomaly detection by proposing the ORCA framework. Integrating supervised contrastive learning with autoencoders, this method constructs a geometric embedding space to generate anomaly scores while enabling template-fitting attribution and uncertainty quantification. Experimental results demonstrate that ORCA significantly enhances sensitivity to new physics searches, accurately recovers signal yields, and effectively characterizes unknown signal features. Consequently, this work establishes a novel paradigm for high-energy physics anomaly detection that successfully combines high performance with robust interpretability, overcoming limitations inherent in previous approaches.
0 citationsRead paper
Recent publications
Latest Papers
Geometric Signatures of Conceptual Reorganization: A Counterfactual Embedding Framework for Detecting Scientific Revolutions
本文通过引入文档嵌入几何和反事实消融框架，提供了一种量化检测科学革命的方法，用以测量单个概念对科学知识组织的影响。
0 citationsRead paper
Latent unified smooth Hamiltonians for excited state chemistry
本文提出一种神经网络架构，通过学习电子态哈密顿量的隐式基表示来统一处理分子系统的基态和激发态问题。
0 citationsRead paper
Panda Diplomacy: Foundation Model Pre-training across Particle Imaging Detectors for High Energy and Nuclear Physics
本文通过点云自蒸馏框架解决了粒子和核物理中基础模型跨传感器复用的问题，提高了模型在不同探测器上的性能。
0 citationsRead paper
Unknown Unknowns: Model Misspecification in Machine Learning for Physics
This study addresses the risks of unknown unknowns arising from model misspecification in physical inverse problems by proposing an iterative diagnosis and mitigation framework. Treating misspecification as an opportunity for discovery, this work establishes a closed-loop detection-mitigation analytical paradigm that integrates complementary diagnostics, iterative updating, and robustness analysis strategies. Consequently, this research develops a systematic methodology for managing unknown unknowns, effectively enhancing model robustness against unforeseen biases and significantly improving the reliability of physical measurements. Ultimately, the proposed framework provides a novel safety assurance mechanism for solving complex inverse problems, ensuring greater confidence in computational reconstructions where model fidelity cannot be fully guaranteed a priori.
0 citationsRead paper
Contrastive Learning for Interpretable Anomaly Detection at Collider Experiments
This study addresses the challenges of score interpretability and strong energy dependence in collider anomaly detection by proposing the ORCA framework. Integrating supervised contrastive learning with autoencoders, this method constructs a geometric embedding space to generate anomaly scores while enabling template-fitting attribution and uncertainty quantification. Experimental results demonstrate that ORCA significantly enhances sensitivity to new physics searches, accurately recovers signal yields, and effectively characterizes unknown signal features. Consequently, this work establishes a novel paradigm for high-energy physics anomaly detection that successfully combines high performance with robust interpretability, overcoming limitations inherent in previous approaches.
0 citationsRead paper