M.N.M. van Lieshout and C. Lu's contribution to the Discussion of 'the Discussion Meeting on Probabilistic and statistical aspects of machine learning'.

The article discusses the work of Professors Li, Fearnhead, Fryzlewics, and Wang on using classic test statistics for change-point detection as a neural network-based classifier and developing improved offline detection algorithms for historical, labelled data. The authors raise the question of whether this approach could be adapted for detecting changes in intensity in a spatio-temporal point pattern, where complex changes in inter-point interaction occur and no known model or labelled historical data is available. The article also mentions other machine learning approaches that have been used in spatio-temporal statistical practice, such as random forest importance scores for variable selection and training neural networks on simulated data for distinguishing spatial structural differences. The funding for the research is acknowledged, and the authors' contact information is provided. [Extracted from the article]