Difficulty 1: Data are too large to be handled.
Many future sky survey projects are in Petabyte (PB; 1024 TB) level.
With Machine Learning, we can greatly reduce the processing time of massive data.
Difficulty 2: Features are too complicated.
With the flood of observational data, not only the quantity but also the information is increasing as well. It is not easy to analyze the data with tranditional methods.
E.g. There are 12,000 parameters in AKARI's data analysis.
Machine learning can handle complicated features effectively.
Neural Network
Random Forrest
Generative adversarial network(GAN)