AttenGW: A Lightweight Attention-Based Multi-Detector Gravitational-Wave Detection Pipeline

TL;DR

AttenGW is a novel attention-based multi-detector gravitational-wave detection pipeline that improves false-positive rates and maintains stable performance on real LIGO data, offering an efficient alternative to graph-based methods.

Contribution

The paper introduces AttenGW, a lightweight attention-based model with a new aggregation scheme for gravitational-wave detection, outperforming traditional graph-based approaches.

Findings

Reduces false-positive rate compared to graph-based methods

Ensemble of three models matches six-model ensemble performance

Maintains stable detection performance on non-Gaussian noise backgrounds

Abstract

We present AttenGW, an attention-based multi-detector gravitational-wave detection model and accompanying software stack designed for analysis of real LIGO data. AttenGW combines a per-detector hierarchical dilated convolutional network with an attention-based aggregation module that enforces cross-detector coherence, providing an alternative to graph-based aggregation schemes used in previous work. The pipeline adopts a LIGO-style preprocessing and data-loading workflow based on GWOSC time series, with standard whitening and filtering, and is released as a documented Python/PyTorch package. We benchmark AttenGW using simulated injections to estimate sensitive volume and on real O3 data, focusing on the February 2020 segment previously used to evaluate a spatiotemporal graph ensemble. On this month of data, a single AttenGW model reduces the false-positive rate relative to a single graph-based detector by a factor of a few, and an ensemble of three AttenGW models matches the performance of the corresponding six-model ensemble. Injection studies on real LIGO noise further indicate that attention-based aggregation yields stable performance on non-Gaussian backgrounds.