A lot of other posts here are biased to recent papers, and papers that had "a big impact", but miss a lot of foundations. I think this reddit post on the most foundational ML papers gives a lot more balanced overview: https://www.reddit.com/r/MachineLearning/comments/zetvmd/d_i...