标题： 迁移学习使μ子断层扫描能够对材料Z进行分类 显示英文标题

标题： Transfer learning empowers material Z classification with muon tomography

摘要： 宇宙射线缪子源在与不同原子序数（Z值）的材料相互作用时表现出不同的散射角分布，有助于识别各种Z类材料，特别是放射性高Z核元素。 大多数传统的识别方法都基于复杂的缪子事件重建和轨迹拟合过程。 监督式机器学习方法提供了一些改进，但严重依赖于目标材料的先验知识，显著限制了其在检测隐藏材料中的实际应用。 在本工作中，首次将迁移学习引入缪子断层扫描领域。 我们提出了两种轻量级神经网络模型，用于微调和对抗性迁移学习，利用裸材料的缪子断层扫描数据来预测涂层材料的Z类。 通过采用逆累积分布函数方法，可以从有限的数据中获得更准确的散射角分布，与基于传统随机采样的训练相比，预测准确率提高了近4%。 当应用于具有有限标记甚至无标记缪子断层扫描数据的涂层材料时，所提出的方法实现了超过96%的整体预测准确率，其中高Z材料接近99%。 模拟结果表明，与没有迁移的直接预测相比，迁移学习将预测准确率提高了约10%。 这项研究证明了迁移学习在克服与有限标记/无标记数据相关的物理挑战方面的有效性，突显了迁移学习在缪子断层扫描领域中的潜在前景。 显示英文摘要

摘要： Cosmic-ray muon sources exhibit distinct scattering angle distributions when interacting with materials of different atomic numbers (Z values), facilitating the identification of various Z-class materials, particularly those radioactive high-Z nuclear elements. Most of the traditional identification methods are based on complex muon event reconstruction and trajectory fitting processes. Supervised machine learning methods offer some improvement but rely heavily on prior knowledge of target materials, significantly limiting their practical applicability in detecting concealed materials. For the first time, transfer learning is introduced into the field of muon tomography in this work. We propose two lightweight neural network models for fine-tuning and adversarial transfer learning, utilizing muon tomography data of bare materials to predict the Z-class of coated materials. By employing the inverse cumulative distribution function method, more accurate scattering angle distributions could be obtained from limited data, leading to an improvement by nearly 4\% in prediction accuracy compared with the traditional random sampling based training. When applied to coated materials with limited labeled or even unlabeled muon tomography data, the proposed method achieves an overall prediction accuracy exceeding 96\%, with high-Z materials reaching nearly 99\%. Simulation results indicate that transfer learning improves prediction accuracy by approximately 10\% compared to direct prediction without transfer. This study demonstrates the effectiveness of transfer learning in overcoming the physical challenges associated with limited labeled/unlabeled data, highlights the promising potential of transfer learning in the field of muon tomography.