arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2106.03741astro-ph.IMcs.LGgr-qc
收藏

深度学习引力波训练策略

Training Strategies for Deep Learning Gravitational-Wave Searches

Max-Planck-Institut für Gravitationsphysik, Albert-Einstein-Institut(马克斯·普朗克引力物理研究所(阿尔伯特·爱因斯坦研究所));Leibniz Universität Hannover(莱布尼茨汉诺威大学);Friedrich-Schiller-Universität Jena(弗里德里希·席勒耶拿大学);Michael Stifel Center Jena(耶拿迈克尔·斯蒂费尔中心)

Marlin B. Schäfer, Ondřej Zelenka, Alexander H. Nitz, Frank Ohme, Bernd Brügmann

摘要:

紧凑的双星系统发出的引力辐射,是潜在的检测到目前地球上的探测器。从仪器的背景噪声中提取这些信号是一个复杂的问题,并且大多数当前搜索的计算成本取决于源模型的复杂性。深度学习可能能够找到当前算法达到计算极限的信号。在这里,我们将我们的分析限制在来自非旋转二元黑洞的信号上,并系统地测试了将训练数据提交给网络的不同策略。为了评估训练策略的影响,我们重新分析了第一批发布的网络,并将其直接与等效的匹配过滤器搜索进行比较。我们发现,深度学习算法可以将低信噪比(SNR)信号推广到高SNR信号,但反之亦然。因此,在训练过程中提供高SNR信号是没有好处的,并且在早期提供低SNR样本时实现最快收敛。在测试过程中,我们发现当需要虚警概率$<10^{-3}$时,网络有时无法恢复任何信号。我们通过在训练后应用我们称为unbounded Softmax replacement(USR)的修改来解决这个限制。通过这种改变,我们发现机器学习搜索保留了匹配过滤器搜索的灵敏度的$\geq 91.5\%$,错误警报率为每月1次。

英文摘要:

Compact binary systems emit gravitational radiation which is potentially detectable by current Earth bound detectors. Extracting these signals from the instruments' background noise is a complex problem and the computational cost of most current searches depends on the complexity of the source model. Deep learning may be capable of finding signals where current algorithms hit computational limits. Here we restrict our analysis to signals from non-spinning binary black holes and systematically test different strategies by which training data is presented to the networks. To assess the impact of the training strategies, we re-analyze the first published networks and directly compare them to an equivalent matched-filter search. We find that the deep learning algorithms can generalize low signal-to-noise ratio (SNR) signals to high SNR ones but not vice versa. As such, it is not beneficial to provide high SNR signals during training, and fastest convergence is achieved when low SNR samples are provided early on. During testing we found that the networks are sometimes unable to recover any signals when a false alarm probability $<10^{-3}$ is required. We resolve this restriction by applying a modification we call unbounded Softmax replacement (USR) after training. With this alteration we find that the machine learning search retains $\geq 91.5\%$ of the sensitivity of the matched-filter search down to a false-alarm rate of 1 per month.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑