RESOLUTION:
We adopt the number of publications under fields Machine Learning and Deep Learning on Arxiv.
We consider as crashed if on the last months of the year (november, december) there is a drop of over 50% on average monthly publications compared to the first part of the year (january, february, march), corrected for the christmas holidays (considering average only before December 20).

Otherwise, we consider it not crashed.