标题： 利用深度学习探测、提取和估算极端质量比并合的信号 显示英文标题

标题： The detection, extraction and parameter estimation of extreme-mass-ratio inspirals with deep learning

摘要： 空间引力波探测器的主要目标之一是探测和分析极端质量比并合（EMRIs）。 由于EMRI信号复杂且持续时间长，并且其固有微弱的特性，这一任务面临重大挑战。 在这封信中，我们介绍了一种用于空间引力波探测器的EMRI信号检测的两层卷积神经网络（CNN）方法，在信号噪声比（SNR）从50到100的范围内，在1%的假阳性率（FPR）下实现了96.9%的真正阳性率（TPR）。 特别是，通过使用VGG网络可以直接推断出EMRI的关键内在参数，例如超大质量黑洞（SMBH）的质量和自旋以及轨道的初始偏心率。 可以分别以99%和92%的概率确定SMBH的质量和自旋。 这将大大减少后续贝叶斯参数估计中的参数空间和计算成本。 我们的模型对波形模型的准确性依赖性也很低。 本研究强调了深度学习方法在EMRI数据分析中的潜力，能够实现快速的EMRI信号检测和高效的参数估计。 显示英文摘要

摘要： One of the primary goals of space-borne gravitational wave detectors is to detect and analyze extreme-mass-ratio inspirals (EMRIs). This endeavor presents a significant challenge due to the complex and lengthy EMRI signals, further compounded by their inherently faint nature. In this letter, we introduce a 2-layer Convolutional Neural Network (CNN) approach to detect EMRI signals for space-borne detectors, achieving a true positive rate (TPR) of 96.9 % at a 1 % false positive rate (FPR) for signal-to-noise ratio (SNR) from 50 to 100. Especially, the key intrinsic parameters of EMRIs such as mass and spin of the supermassive black hole (SMBH) and the initial eccentricity of the orbit can be inferred directly by employing a VGG network. The mass and spin of the SMBH can be determined at 99 % and 92 % respectively. This will greatly reduce the parameter spaces and computing cost for the following Bayesian parameter estimation. Our model also has a low dependency on the accuracy of the waveform model. This study underscores the potential of deep learning methods in EMRI data analysis, enabling the rapid detection of EMRI signals and efficient parameter estimation .