“Time series analysis can be used in a wide range of areas,” said professor Choi. “Not only in the financial market for stocks and exchange rates, but for nuclear power plants in predicting component malfunctions, in heavy industry, and even the military.” (image: KobizMedia/ Korea Bizwire)
SEOUL, June 28 (Korea Bizwire) – A Korean research team has developed an artificial intelligence system that can analyze constantly-changing stocks, exchange rates, and housing prices.
A team of scientists from the Ulsan National Institute of Science and Technology, led by professor Choi Jae-sik, developed a ‘Relational Automatic Statistician System’, which is an AI-based system that has the ability to discover mutual factors from changing time series data (data which changes over time).
The team implemented an AI analytic technique into an existing Gaussian model to develop an algorithm that can predict time series data, which often shows atypical behaviors, with improved accuracy.
In fact, upon the team’s experimental application, the new system was able to predict the overall trend of the stock market after the 9/11 attack, where top stocks dropped and then steadily recovered.
The team also explained that applying the system to nuclear power plants would help determine whether specific parts showing unusual symptoms were signs of malfunction or mere changes within the boundary of normal activities.
“Time series analysis can be used in a wide range of areas,” said professor Choi. “Not only in the financial market for stocks and exchange rates, but for nuclear power plants in predicting component malfunctions, in heavy industry, and even the military.”
The results of the study were first announced on June 22 at the International Conference on Machine Learning held in New York.
By Joseph Shin (jss539@koreabizwire.com)
