Abstract:We present a new method to search for long transient gravitational waves signals, like those expected from fast spinning newborn magnetars, in interferometric detector data. Standard search techniques are computationally unfeasible (matched filtering) or very demanding (sub-optimal semi-coherent methods). We explored a different approach by means of machine learning paradigms, to define a fast and inexpensive procedure. We used convolutional neural networks to develop a classifier that is able to discriminate between the presence or the absence of a signal. To complement the classification and enhance its effectiveness, we also developed a denoiser. We studied the performance of both networks with simulated colored noise, according to the design noise curve of LIGO interferometers. We show that the combination of the two models is crucial to increase the chance of detection. Indeed, as we decreased the signal initial amplitude (from $10^{-22}$ down to $10^{-23}$) the classification task became more difficult. In particular, we could not correctly tag signals with an initial amplitude of $2 \times 10^{-23}$ without using the denoiser. By studying the performance of the combined networks, we found a good compromise between the search false alarm rate (2$\%$) and efficiency (90$\%$) for a single interferometer. In addition, we demonstrated that our method is robust with respect to changes in the power law describing the time evolution of the signal frequency. Our results highlight the computationally low cost of this method to generate triggers for long transient signals. The study carried out in this work lays the foundations for further improvements, with the purpose of developing a pipeline able to perform systematic searches of long transient signals.

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? This paper aims to solve the problem of searching for long - transient gravitational - wave signals (such as those emitted by rapidly rotating newly - born magnetars) in gravitational - wave (GW) data. Traditional search techniques, such as matched filtering, are too computationally expensive, and sub - optimal semi - coherent methods can be used but are very resource - intensive. Therefore, the researchers explored a different approach based on the machine - learning paradigm to define a fast and cost - effective procedure. Specifically, the paper proposes the following points: 1. **Develop a convolutional neural network (CNN) classifier**: It is used to distinguish whether there are long - transient gravitational - wave signals. 2. **Develop a denoiser**: To improve the performance of the classifier, the researchers also developed a denoiser to reduce the impact of noise. 3. **Combine the classifier and the denoiser**: By combining these two models, the ability to detect weak signals is improved, especially when the initial amplitude of the signal is low (from $10^{-22}$ to $10^{-23}$). 4. **Evaluate performance**: The researchers tested the performance of these models in a simulated colored - noise environment and showed a good balance between the false - positive rate (2%) and the efficiency (90%) on a single interferometer. 5. **Verify robustness**: It is proved that this method is robust to changes in the power - law description of the signal frequency over time. Overall, this paper proposes a new, low - cost - in - terms - of - computation method that can generate trigger signals for long - transient gravitational - wave signals, thus laying the foundation for a more systematic search for such signals in the future. ### Formula presentation - **Spin - down equation**: \[ \dot{f}_{\text{rot}}=-k f_{\text{rot}}^n \] where $k$ and $n$ are the spin - down constant and the braking index respectively, which are determined by the emission mechanism. - **Relationship between gravitational - wave frequency and rotation frequency**: \[ f(t) = 2 f_{\text{rot}}(t) \] - **Gravitational - wave strain**: \[ h_+(t)=h_0(t)\frac{1 + \cos^2\iota}{2}\cos(2\pi t f(t)) \] \[ h_x(t)=h_0(t)\cos\iota\sin(2\pi t f(t)) \] where $h_0(t)=\frac{4\pi^2 G I f(t)^2}{c^4 d\epsilon}$, $G$ is the gravitational constant, $C$ is the speed of light, $I$ is the moment of inertia, $d$ is the distance, and $\epsilon$ is the ellipticity. - **Pixel signal - to - noise ratio (PSNR)**: \[ \text{PSNR}=\frac{1}{N_{\text{pix}}}\sum_{t, f\neq0}\frac{S_{tf}}{N_{tf}} \] - **Overlap measure $O$**: \[ O = \sqrt{\frac{\sum_{tf}h_{tf}h_{d, tf}}{\sum_{tf}h_{tf}h_{tf}}} \] - **F1 score**: \[ F1=\frac{2\times P\times R}{P + R}=\frac{2\times TP}{2\times TP+FN + FP} \] where $P=\frac{TP}{TP + FP}$, $R=\frac{TP}{TP + FN}$ Through these formulas and methods, the researchers showed how to effectively identify long - transient gravitational - wave signals, thus providing a solid foundation for future research.

Neural network method to search for long transient gravitational waves

Convolutional neural network search for long-duration transient gravitational waves from glitching pulsars

Detection and Parameter Estimation of Gravitational Waves from Binary Neutron-Star Mergers in Real LIGO Data using Deep Learning

Neural network time-series classifiers for gravitational-wave searches in single-detector periods

A novel multi-layer modular approach for real-time fuzzy-identification of gravitational-wave signals

A neural network-based gravitational wave interpolant with applications to low-latency analyses

New approach to search for long transient gravitational waves from inspiraling compact binary systems

Real-Time Detection of Gravitational Waves from Binary Neutron Stars using Artificial Neural Networks

A machine-learning pipeline for real-time detection of gravitational waves from compact binary coalescences

Deep Residual Networks for Gravitational Wave Detection

Convolutional Neural Networks for signal detection in real LIGO data

Real-Time Detection of Unmodelled Gravitational-Wave Transients Using Convolutional Neural Networks

Detecting optical transients using artificial neural networks and reference images from different surveys

SpecGrav -- Detection of Gravitational Waves using Deep Learning

Applying Deep Neural Networks to the Detection and Space Parameter Estimation of Compact Binary Coalescence with a Network of Gravitational Wave Detectors

Search for exotic gravitational wave signals beyond general relativity using deep learning

A new method to distinguish gravitational-wave signals from detector noise transients with Gravity Spy

Improving the background of gravitational-wave searches for core collapse supernovae: a machine learning approach

Machine learning for gravitational-wave detection: surrogate Wiener filtering for the prediction and optimized cancellation of Newtonian noise at Virgo

Exploring gravitational-wave detection and parameter inference using deep learning methods