Stories
Slash Boxes
Comments

SoylentNews is people

posted by Fnord666 on Wednesday April 18 2018, @02:50AM   Printer-friendly
from the data-scientists-are-now-gold-diggers dept.

As humankind continues to stare into the dark abyss of deep space in an eternal quest to understand our origins, new computational tools and technologies are needed at unprecedented scales. Gigantic datasets from advanced high resolution telescopes and huge scientific instrumentation installations are overwhelming classical computational and storage techniques.

This is the key issue with exploring the Universe – it is very, very large. Combining advances in machine learning and high speed data storage are starting to provide hitherto unheard of levels of insight that were previously in the realm of pure science fiction. Using computer systems to infer knowledge from observation isn’t new, but the scale at which we need to examine large data today certainly is.

Because the data are so unwieldy and complicated, new methods need to be devised to quickly annotate features that are important, sifting out valuable signals from all of the noise. Nothing is probably more difficult than finding the signal of an “echo” observed from the “sound” a pair of black holes colliding billions of light years away from Earth. This was the premise of the algorithms needed to make sense of the data from the Laser Interferometer Gravitational-Wave Observatory (LIGO) project. They need to filter out a vast array of noise from the real “proton sized” signal, it is an intrinsically computationally intensive process, the main reason being just the sheer size and noiseiness of the captured data.

The team has now published two papers on their methods to use AI to find gravitational waves. First up was the initial paper, “Deep neural networks to enable real-time multimessenger astrophysics” in Physics Review D, in February 2018, with the follow-on Deep Learning for real-time gravitational wave detection and parameter estimation: Results with Advanced LIGO data in Physics Letters B, in March 2018.


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 0) by Anonymous Coward on Wednesday April 18 2018, @10:27PM

    by Anonymous Coward on Wednesday April 18 2018, @10:27PM (#668745)

    Shaddup Juan