arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2103.03557gr-qcastro-ph.IM
收藏

引力波探测中卷积神经网络对瞬态噪声的响应

The response of the Convolutional Neural Network to the transient noise in Gravitational wave detection

Chao Zhan, Mingzhen Jia, Cunliang Ma, Zhongliang Lu, Wenbin Lin

摘要:

近年来,许多工作研究了卷积神经网络用于引力波探测。然而,很少有人关注瞬态噪声是否会触发CNN模型。在本文中,我们研究了正弦高斯毛刺,高斯毛刺和振铃毛刺的训练卷积神经网络分类器的响应。我们发现,该网络是鲁棒的正弦高斯和高斯毛刺,其虚警概率接近的LIGO类噪声,在对比的情况下,衰荡毛刺,其中的虚警概率远远大于LIGO类噪声。我们还研究了不同频率的毛刺的响应。我们发现当毛刺的频率落在训练的GW信号的频率时,毛刺的虚警概率将远大于LIGO类噪声的虚警概率,并且毛刺被误判为GW信号的概率甚至可以超过30%。

英文摘要:

In recent years, much work have studied the use of convolutional neural networks for gravitational-wave detection. However little work pay attention to whether the transient noise can trigger the CNN model or not. In this paper, we study the responses of the sine-Gaussian glitches, the Gaussian glitches and the ring-down glitches in the trained convolutional neural network classifier. We find that the network is robust to the sine-Gaussian and Gaussian glitches, whose false alarm probabilities are close to that of the LIGO-like noises, in contrast to the case of the ring-down glitches, in which the false alarm probability is far larger than that of the LIGO-like noises. We also investigate the responses of the glitches with different frequency. We find that when the frequency of the glitches falls in that of the trained GW signals, the false alarm probability of the glitches will be much larger than that of the LIGO-like noises, and the probability of the glitches being misjudged as the GW signals may even exceed 30%.

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。

↑