Gravity Spy: integrating advanced LIGO detector characterization, machine learning, and citizen science.

Area of Science:

• Astronomy and Astrophysics
• Gravitational Wave Detection
• Data Analysis and Signal Processing

Background:

• The Laser Interferometer Gravitational-Wave Observatory (LIGO) enables a new era of astronomy through direct gravitational wave detection.
• LIGO's extreme sensitivity makes it susceptible to instrumental and environmental noise, particularly transient, non-Gaussian events called glitches.
• Glitches can obscure or mimic genuine gravitational wave signals, hindering the accurate detection rates predicted by LIGO's design sensitivity.

Purpose of the Study:

• To develop an efficient and accurate method for categorizing the vast number of glitches recorded by LIGO detectors.
• To leverage crowdsourcing and machine learning to aid in glitch characterization, a task challenging for scientists alone.
• To improve the identification and potential elimination of glitches, thereby enhancing the rate and accuracy of gravitational wave observations.

Main Methods:

• An innovative project combining crowdsourcing via the Zooniverse platform with machine learning algorithms.
• Public volunteers categorize time-frequency images of glitches into predefined morphological classes and identify new ones.
• Machine learning algorithms are trained on human-classified examples to categorize glitches, with a combined approach enhancing performance.

Main Results:

• Demonstration of the combined crowdsourcing and machine learning method using a subset of data from LIGO's first observing run.
• The integrated approach aims to improve the efficiency and accuracy of individual classification methods.
• The resulting classification and characterization of glitches provide valuable data for LIGO scientists.

Conclusions:

• The combined crowdsourcing and machine learning approach offers a scalable solution for categorizing LIGO glitches.
• This method aids scientists in identifying glitch origins, facilitating their removal from data or the detector.
• Improved glitch characterization is crucial for maximizing LIGO's scientific output and achieving its design sensitivity goals.