Abstract
Ultrashort-pulse laser processing is well-suited for microprocessing as it can suppress heat effects around
the processed area. Achieving high-quality processing requires the optimization of parameters, which
requires much labor and time. An effective way to avoid this is through efficient data collection and application
of machine learning. The effects of laser parameters such as pulse duration and repetition rate
on the laser processing are reviewed. We present the automatic parameter-variable ultrashort-pulse laser
processing system to help collect large amount of data which can be used for supervised machine learning.
We review the studies on application of machine learning to prediction of 3-dimensional shape of
processed craters.
