Fast sky localization of gravitational waves using deep learning seeded importance sampling

Kolmus, Alex; Baltus, G.; Janquart, Justin; Laarhoven, Twan van; Caudill, S.; Heskes, Tom

doi:10.1103/physrevd.106.023032

Cited by 7 publications

(7 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The main advantage of ML techniques is their rapidity because most of the computations are made during the training stage. A widely used ML method for pattern recognition is based on convolutional neural networks (CNNs) [51], in the context of GW it has been applied to different tasks such as CBC identification [52][53][54][55][56], burst detection [57][58][59][60], sky localization [61][62][63], glitch classification [64,65] and synthetic data generation [66,67]. See [68] for a review on this topic.…”

Section: Introductionmentioning

confidence: 99%

Convolutional neural network for gravitational-wave early alert: Going down in frequency

et al. 2022

Self Cite

View full text Add to dashboard Cite

We present here the latest development of a machine-learning pipeline for premerger alerts from gravitational waves coming from binary neutron stars (BNSs). This work starts from the convolutional neural networks introduced in [Baltus et al., Phys. Rev. D 103, 102003 (2021)] that searched for the early inspirals in simulated Gaussian noise colored with the design-sensitivity power-spectral density of LIGO. Our new network is able to search for any BNS with a chirp mass between 1 and 3 M ⊙ , it can take into account all the detectors available, and it can see the events even earlier than the previous one. We study the performance of our method in three different scenarios: colored Gaussian noise based on the O3 sensitivity, real O3 noise, colored Gaussian noise based on the predicted O4 sensitivity. We show that our network performs almost as well in non-Gaussian noise as in Gaussian noise: our method is robust with respect to glitches and artifacts present in real noise. Although it would not have been able to trigger on the BNSs detected during O3 because their signal-to-noise ratio was too weak, we expect our network to find around 3 BNSs during O4 with a time before the merger between 3 and 88 s in advance.

show abstract

Section: Introductionmentioning

confidence: 99%

Convolutional neural network for gravitational-wave early alert: Going down in frequency

et al. 2022

Self Cite

View full text Add to dashboard Cite

show abstract

“…possible avenue is applying importance sampling after the normalizing flow [45,61]. However, such methods can be tricky, and additional modifications to our network could be needed.…”

mentioning

confidence: 99%

Normalizing Flows as an Avenue to Studying Overlapping Gravitational Wave Signals

et al. 2023

Self Cite

View full text Add to dashboard Cite

Because of its speed after training, machine learning is often envisaged as a solution to a manifold of the issues faced in gravitational-wave astronomy. Demonstrations have been given for various applications in gravitational-wave data analysis. In this Letter, we focus on a challenging problem faced by thirdgeneration detectors: parameter inference for overlapping signals. Because of the high detection rate and increased duration of the signals, they will start to overlap, possibly making traditional parameter inference techniques difficult to use. Here, we show a proof-of-concept application of normalizing flows to perform parameter estimation on overlapped binary black hole systems.

show abstract

“…Importance sampling requires evaluation of p(d|θ)p(θ) rather than the normalized posterior. The Bayesian evidence can then be estimated from the normalization of 2 A similar approach using convolutional networks to parametrize Gaussian and von Mises proposals was used to estimate the sky position alone [33] Using the normalizing flow proposal (as we do here) significantly improves the flexiblity of the conditional density estimator and enables inference of all parameters.…”

mentioning

confidence: 99%

Neural Importance Sampling for Rapid and Reliable Gravitational-Wave Inference

Dax¹,

Green²,

Gair³

et al. 2022

Preprint

View full text Add to dashboard Cite

We combine amortized neural posterior estimation with importance sampling for fast and accurate gravitational-wave inference. We first generate a rapid proposal for the Bayesian posterior using neural networks, and then attach importance weights based on the underlying likelihood and prior. This provides (1) a corrected posterior free from network inaccuracies, (2) a performance diagnostic (the sample efficiency) for assessing the proposal and identifying failure cases, and (3) an unbiased estimate of the Bayesian evidence. By establishing this independent verification and correction mechanism we address some of the most frequent criticisms against deep learning for scientific inference. We carry out a large study analyzing 42 binary black hole mergers observed by LIGO and Virgo with the SEOBNRv4PHM and IMRPhenomXPHM waveform models. This shows a median sample efficiency of ≈ 10% (two orders-of-magnitude better than standard samplers) as well as a ten-fold reduction in the statistical uncertainty in the log evidence. Given these advantages, we expect a significant impact on gravitational-wave inference, and for this approach to serve as a paradigm for harnessing deep learning methods in scientific applications.

show abstract

Fast sky localization of gravitational waves using deep learning seeded importance sampling

Cited by 7 publications

References 33 publications

Convolutional neural network for gravitational-wave early alert: Going down in frequency

Convolutional neural network for gravitational-wave early alert: Going down in frequency

Normalizing Flows as an Avenue to Studying Overlapping Gravitational Wave Signals

Neural Importance Sampling for Rapid and Reliable Gravitational-Wave Inference

Contact Info

Product

Resources

About