用视觉Transformer分析天文光变曲线图像，提升系外行星识别准确率
Exoplanet Classification through Vision Transformers with Temporal Image Analysis
- 将光变曲线转为时序图像，输入ViT模型捕捉时间依赖关系
- 在开普勒数据上达到89.46%召回率与85.09%精确率
- 适合关注自动化天体分类与深度学习应用的研究者
系外行星分类是天文学中的长期挑战，传统方法耗时耗力。本文提出一种新方法，将来自NASA开普勒任务的原始光变曲线数据，通过格拉米安角差场（GAF）和递归图（RP）技术转化为图像，输入视觉变换器（ViT）模型，利用其捕捉复杂时间依赖性的能力。采用五折交叉验证评估性能，以减少偏差。结果表明，递归图（RP）优于格拉米安角场（GAF），ViT模型实现89.46%召回率和85.09%精确率，显著提升系外行星掩星事件的识别能力。尽管使用欠采样缓解类别不平衡问题，但数据集规模仍受限。研究强调需进一步优化模型架构，以提升自动化水平、性能与泛化能力。
原文摘要 · Abstract (English)
The classification of exoplanets has been a longstanding challenge in astronomy, requiring significant computational and observational resources. Traditional methods demand substantial effort, time, and cost, highlighting the need for advanced machine learning techniques to enhance classification efficiency. In this study, we propose a methodology that transforms raw light curve data from NASA's Kepler mission into Gramian Angular Fields (GAFs) and Recurrence Plots (RPs) using the Gramian Angular Difference Field and recurrence plot techniques. These transformed images serve as inputs to the Vision Transformer (ViT) model, leveraging its ability to capture intricate temporal dependencies. We assess the performance of the model through recall, precision, and F1 score metrics, using a 5-fold cross-validation approach to obtain a robust estimate of the model's performance and reduce evaluation bias. Our comparative analysis reveals that RPs outperform GAFs, with the ViT model achieving an 89.46$\%$ recall and an 85.09$\%$ precision rate, demonstrating its significant capability in accurately identifying exoplanetary transits. Despite using under-sampling techniques to address class imbalance, dataset size reduction remains a limitation. This study underscores the importance of further research into optimizing model architectures to enhance automation, performance, and generalization of the model.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。