Earthquake is among the most dreadful natural calamities the human race has come across. It has the potential to destroy almost everything. Thus sensing its approach proves to be very important in order to take necessary steps. An artificial intelligence system is developed by a scientist to successfully foresee earthquakes, a progress that may assist to organize for natural disasters and thus save lives.
The study recognized a hidden signal clueing to earthquakes and utilized this “fingerprint” to skill a machine learning algorithm to envisage future earthquakes. The team of researchers from the Boston University, the United States, and the University of Cambridge, the United Kingdom, studied the relations among earthquakes, faults, and precursor quakes, with the hopefulness of developing a technique to foresee earthquakes.
With the use of a lab-based system that imitates actual earthquakes, the research team utilized machine learning methods to analyze the acoustic signals approaching from the fault as it stimulated and explore for patterns. The team utilized steel blocks to thoroughly imitate the physical forces at toil in an actual earthquake and also logs the seismic sounds and signals that are released.
Machine learning was then utilized to exploit the association between the acoustic signal released from the fault and how near it is toward weakening. The machine learning algorithm was capable of recognizing a specific sound pattern, earlier deemed to be nothing more than sound that occurs long prior to an earthquake, according to the researchers.
The features of this sound pattern can be utilized to provide an accurate evaluation of the pressure on the fault and to assess the time lasting prior to failure that becomes more and more specific as failure nears, they said.
Cambridge University’s Colin Humphreys, said, “For the first time, machine learning has been utilized to analyze the acoustic information to forecast when an earthquake will take place, long before it does, thus making available plenty of cautionary time—it is implausible what machine learning can do.”
Machine learning allows the scrutiny of datasets too huge to manage manually and explores data in an impartial manner that lets discoveries to be accomplished, according to researchers.