User:EnIRtpf09b/sandbox/Gravity Spy

Gravity Spy is a Zooniverse citizen science project in which volunteers classify data from the LIGO and Virgo gravitational-wave observatories to help scientists train machine-learning algorithms to identify noise sources known as glitches and eliminate them. It was launched by a team of LIGO researchers from the Center for Interdisciplinary Exploration and Research in Astronomy (CIERA) and machine-learning experts from the Northwestern University, researchers at CalTech, crowd-sourced science researchers at Syracuse University and Zooniverse web developers.

History
During the project's beta testing stage, 1400 volunteers registered on the project and made a total of 45,000 glitch classifications and discovered two new classes of glitches on the project. After beta-testing, the project was launched as an official Zooniverse project on October 12, 2016. In October 2018, data from the Virgo interferometer was added to the project along with a supplementary website for Gravity Spy volunteers to use known as Gravity Spy Tools.

As of April 2022, 28,853 registered volunteers participated in the project together making more than 6,000,000 classifications with 13 scientific papers and articles published by the project team.

Gravity Spy 2.0
On 5 October 2021, the Gravity Spy team were awarded a three-year research grant by the National Science Foundation (NSF) to continue with the next phase of Gravity Spy, Gravity Spy 2.0. The primary goal of the project will be to identify the causes of glitches by searching for correlations between glitches in the main and auxiliary channels. The team also plans to create Beginner, Intermediate, and Advanced workflows that support different modes of interaction at each stage, similar to the workflow system in Gravity Spy.

Non-Gravity Spy Project: GWitchHunters
On 16 November 2021, a project called GWitchHunters was launched by a team of researchers from the Virgo Collaboration and the Research Infrastructures for Citizens in Europe (REINFORCE) under a grant by the Community Research and Development Information Service (CORDIS) of the European Commission. GWitchHunters is similar to Gravity Spy 2.0 with volunteers also searching for correlations between the main and auxiliary channels. However, unlike Gravity Spy 2.0, GWitchHunters uses data only from the Virgo interferometer.

Classification
In Gravity Spy, volunteers are presented with spectrograms and are asked to identify glitches into its classes based on the characteristics of the glitch. Volunteers can also see different time durations of the same glitch. This feature is helpful for looking at glitches that happen over longer periods of time. After doing a certain number of classifications, registered volunteers get access to more advanced workflows that have more glitch classes and more complicated subjects. There are a total of 7 such workflows each named after astrophysical signals that LIGO may be able to detect.

Literature

 * Gravity Spy: Integrating Advanced LIGO Detector Characterization, Machine Learning, and Citizen Science
 * Deep multi-view models for glitch classification
 * Gravity Spy: Humans, machines and the future of citizen science
 * Blending machine and human learning processes
 * Recruiting messages matter: Message strategies to attract citizen scientists
 * Machine learning for Gravity Spy: Glitch classification and dataset.
 * Folksonomies to support coordination and coordination of folksonomies.
 * Appealing to different motivations in a message to recruit citizen scientists: results of a field experiment
 * Did they login? Patterns of anonymous contributions to online communities
 * DIRECT: Deep DIscRiminative Embedding for ClusTering of LIGO Data
 * Scaffolding training with machine learning: An experiment on participant learning in an on-line production community
 * Discovering features in gravitational-wave data through detector characterization, citizen science and machine learning
 * Shifting Forms of Engagement: Volunteer Learning in Online Citizen Science. Proceedings of the ACM on Human-Computer Interaction