Well, reinforcement learning should also be given a mention and revolutions there (e.g AlphaGo to AlphaZero) were much more recent