Machine learning helps scientists peer (one second) into the future

The past can be a fixed and unchanging point, but with the help of machine learning, the future can sometimes be more easily guess.

Using a new type of machine learning method called next generation tank computingOhio State University researchers recently found a new way to predict the behavior of chaotic spatiotemporal systems – such as changes in Earth’s weather – that are particularly complex for scientists to predict.

The study, published today in the journal Chaos: an interdisciplinary journal of nonlinear sciencesuses a new, highly efficient algorithm that, when combined with next-generation reservoir computing, can learn spatiotemporal chaotic systems in a fraction of the time of other machine learning algorithms.

The researchers tested their algorithm on a complex problem that has been studied many times in the past: predicting the behavior of an atmospheric weather model. Compared to traditional machine learning algorithms that can solve the same tasks, the Ohio State team’s algorithm is more accurate and uses 400 to 1,250 times less training data to make better predictions than his counterpart. Their method is also less computationally expensive; While solving complex computing problems previously required a supercomputer, they used a laptop running Windows 10 to make predictions in about a split second, about 240,000 times faster than traditional machine learning algorithms. Wendson Barbosa

“This is very exciting, as we believe this is a substantial step forward in terms of data processing efficiency and prediction accuracy in the field of machine learning,” said Wendson De Sa Barbosa, lead author and postdoctoral researcher in physics at Ohio State. He said learning to predict these extremely chaotic systems is a “great physical challenge” and understanding them could pave the way for new scientific discoveries and breakthroughs.

“Modern machine learning algorithms are particularly well suited to predict dynamical systems by learning their underlying physical rules using historical data,” said By Sa Barbosa. “Once you have enough data and computing power, you can make predictions with machine learning models on any complex real-world system.” Such systems can include any physical process, from the swinging of a clock’s pendulum to disturbances in power grids.

Even heart cells display chaotic spatial patterns when they oscillate at an abnormally higher frequency than a normal heartbeat, De Sa Barbosa said. This means that this research could one day be used to better understand the control and interpretation of heart disease, as well as a host of other “real world” issues.

“If one knows the equations that accurately describe how these unique processes for a system will evolve, then its behavior could be reproduced and predicted,” he said. Simple movements, like the swing position of a clock, can be predicted easily using only its current position and speed. Yet more complex systems, like the Earth’s climate, are much more difficult to predict due to the number of variables that actively dictate its chaotic behavior.

To make accurate predictions of the entire system, scientists would need to have precise information about each of these variables, and the model equations that describe how these many variables are related, which is utterly impossible, De said. His Barbosa. But with their machine learning algorithm, the nearly 500,000 historical training data points used in previous work for the atmospheric weather example used in this study could be reduced to just 400, while still achieving the same accuracy. or better accuracy.

Going forward, De Sa Barbosa aims to further research using their algorithm to possibly speed up spatiotemporal simulations, he said.

“We live in a world we still know so little about, so it’s important to recognize these highly dynamic systems and learn how to predict them more effectively.”

The co-author of the study was Daniel J. Gauthier, a physics professor at Ohio State. Their work was supported by the Air Force Office of Scientific Research.


#Machine #learning #helps #scientists #peer #future

Leave a Comment

Your email address will not be published. Required fields are marked *