Enhancing Gravitational-Wave Detection: A Machine Learning Pipeline Combination Approach with Robust Uncertainty Quantification
增强引力波探测:结合稳健不确定性量化的机器学习流水线组合方法
Gravitational-wave data from advanced-era interferometric detectors consists of background Gaussian noise, frequent transient artifacts, and rare astrophysical signals. Multiple search algorithms exist to detect the signals from compact binary coalescences, but their varying performance complicates interpretation. We present a machine-learning-driven approach that combines results from individual pipelines and utilizes conformal prediction to provide robust, calibrated uncertainty quantification. Using simulations, we demonstrate improved detection efficiency and apply our model to GWTC-3, enhancing confidence in multipipeline detections, such as the subthreshold binary neutron star candidate GW200311_103121.
展开 ▾首次将机器学习流水线组合与共形预测结合,克服了传统“取最大pastro/IFAR”的局限,能在无分布假设下对候选体给出可靠置信度,显著增强亚阈值信号(如双中子星候选体GW200311_103121)的探测信心。
既往引力波瞬态目录合并多流水线结果时常取最大pastro或IFAR,虽然简单但未能充分利用流水线间的互补信息。Banagiri+ 2023曾尝试统一多流水线的pastro,但需依赖精确的信号与噪声分布模型。本文提出用机器学习学习流水线间的关联,并引入共形预测(Ashton+ 2024曾将其用于单流水线校准)提供稳健的小样本不确定性量化,从而在无严格模型假设下提升组合探测效率。展望未来,该方法可扩展至多类别源分类、纳入更多流水线特征与探测器状态信息,若能配合更真实的模拟数据并保证交换性,有望成为多流水线联合分析及实时预警的标准化框架。
预印本 2025-04-24 · 刊出 2026-01-08 · 收录 2026-10-12