Machine Learning in Space
Machine learning has always been a developing field. It has assisted us in nearly every function. From cameras of security to home chores and entertainment even, the machine learning can be claimed as a pinnacle field of technology.
But what happens when machine learning is involved in space sciences? Well, we get interesting and new discoveries. The modern discovery of two exo-planets is one such remarkable achievement of machine learning. It is now due to machine learning that scientists can discover more about the two planets. This could present a new field of study and research. Not to mention the benefit potential that could be hidden in this…
A machine learning called neural networks is involved in this discovery. The neural network technology has identified these two extra planets in our solar system. After they were discovered and confirmed, the NASA scientists and Google engineers stepped in to take the glory. It was them who finally announced it.
Two New ExoPlanets
You may perceive it to be a wonderful occurrence. The discovery of new exoplanets is surely a wonderful achievement. However, what you don’t know is that it is not uncommon to find exoplanets outside our solar system. In fact, it is so common that scientists have developed a device named as Kepler Space Telescope. This telescope has shown remarkable findings to the researchers. A proof of its functionality is that it has discovered more than 2000 such planets. All of them lie outside our solar system.
So what is really interesting about these two exoplanets? I mean we already have found more than 2000 extra planets outside our solar system…right?
What is interesting is that these two have been found in our own solar system. That will surely mess up all of the written books of the solar system! Depending on the information received from these two planets, we can all be sure that the exoplanets hide potential benefits from their research. These two exoplanets have been named as Kepler - 90i and Kepler - 80g.
Although both are very fascinating, the 90i planet is especially interesting to researchers. The planet may specify itself as a standalone planet. This will include total planets of 8 in our solar system. Another interesting feature of this planet is the levels of heat. The planet is blazing hot. It easily maintains a temperature near 800 degrees Fahrenheit.
The discovery exoplanets is quite common in outer space. The level of common is shared by the neural networks. The neural networks are common in modern technology too. In these networks, special software is involved. This software learns from data. Majority of programs as we use in our tech are preprogrammed. This means they have already been set to do some specific things. A digital thermometer won’t notify why the change in temperature is occurring. It will simply note down the temperatures.
In case of neural networks, the software involved reacts in a specific way to information gathered from a data. We see many examples of these neural networks in our daily life. Facial recognition in security systems, Google language translator, as well as image recognition facility, they all fall under one umbrella of neural network technology.
Neural network will simply take a cat picture and identify the pictures of other cats even if they are entirely different. The system learns on its own.
The astronomers often use the tools like telescope. They commonly utilize the telescopes to search for the planets and exo – planets. A Kepler telescope collects a data in form of light curves. The light curves actually measure the changes in star light. Every star has a planet orbiting it. When a planet is passing, it results in specific light changes. These light changes of stars are recorded in form of signals.
Nearly 15,000 signals had been taken. Out of these, 3600 were genuine signals. The rest were simply false positives.
Astronomers decided to involve artificial intelligence into it. The neural network was fed with this signal. All 15000 data signals were fed to it. The AI was taught to differentiate between genuine planet light curves and the false positives. The neural network learned it pretty well.
The research was held in the University of Texas. The researchers there decided to place the newly created and trained artificial intelligence system in the Kepler Telescope. The original training data set of Artificial intelligence was different. However, it has the ability to learn on its own.
The research was conducted on 670 signals that the Kepler – telescope managed to capture. The data set was tricky. This meant that the procedures weren’t ended here. In the weakest of signals, the patterns were found. These patterns indicated that there was something like two extra planets. As researchers focused their effort onto understanding these 670 signals, they found that the two new worlds of planets were waiting for them.
One of the leading researchers expressed his faith on artificial intelligence and neural networks. He said that where the data becomes too big for human understanding, the machine learning takes charge. Like it did here, machine learning can be the new future of technology involved in astronomy.
The game doesn’t end here. The proof of two new worlds was hidden here. While this is quite a big achievement, the research won’t stop. There are literally so many of worlds out there, awaiting our senses to grow stronger. The researchers are passionate about finding new worlds and discovering the new benefits they could bring.
While artificial intelligence has been used many times before in astronomy, it was first time that it was designed this way. The neural network designed specifically for this purpose was just the start of something big.