Machine Astronomers Are Looking For The Unknowns In The Universe
Scientists are developing machine astronomers to help them find the unknowns in the universe. This Is because data from the many telescopes are too huge for humans to analyze and study. With the help of machines, not only will they cut down the time to analyze and categorize the data but the machine astronomers can also find unexpected unknowns.
According to an article in The Conversation, most of the spectacular discoveries in Astronomy are unexpected. Many of the telescopes used in Astronomy are developed and used to find what people know to be unknown. However, people have overlooked the potential of the telescopes to observe and detect the real unknowns. These real unknowns when discovered are what make true breakthroughs in Astronomy.
An example of a real unknown is about the discovery of pulsars, according to Space.com. Jocelyn Bell Burnell, a young PhD student in the U.K. in the 1960s, discovered the pulsars by laboriously analyzing data and using a different parameter to explain the inconsistencies in her telescopic readings. If she had not accidentally observed the inconsistencies and investigated them, people would never know anything about pulsars.
However, unlike Jocelyn Burnell who poured through data to discover pulsars, the data now being transmitted by telescopes are too huge for any human to analyze and study. One telescope alone, CSIRO's ASKAP Telescope, is expected to transmit at least 80 petabytes a year. A petabyte is equivalent to 1,048,576 gigabytes.
With the huge amount of data to be expected, scientists are developing machine astronomers to help them. Machine astronomers are just computer machines trained through machine learning techniques and exposed to theory-based simulations to do the work for human astronomers. The machine astronomers will be faster in analyzing and categorizing data from the telescopes and they can be trained to find real unknowns.
There is already a machine astronomer being developed. Project WTF (Widefield ouTlier Finder) is a machine designed to shift through petabytes of data looking for unknowns. By training the WTF through machine learning techniques and exposing it through theory-based simulations, it can find both known unknowns and real unknowns in the universe.
Furthermore, the scientists are not limiting the parameters of WTF so it can go truly and literally analyze and categorize the data in any way it wants. Currently, the team of developers of Project WTF is still training the machine and improving its "brain" by exposing it to different kinds of data mining techniques. WTF is also slowly being trained to go beyond the expected parameters in Astronomy to find the real unknowns in the universe.