What's disheartening, is that all the charts are in log scale for "Computation (Hardware Burden)". These days it feels like any non-trivial achievement in the f

What's disheartening, is that all the charts are in log scale for "Computation (Hardware Burden)". These days it feels like any non-trivial achievement in the field of ML requires resources of the state actors. Maybe that's the sign of ML becoming a mature field like Physics, where collective efforts like CERN and ITER are required, and all low hanging fruits have been picked. Bot it also goes in contrast with how Mathematics and Statistics research has been made before. In these fields a paper usually has only one to three authors, and one man can make a difference. In ML, it's probably not the case anymore.