17 December 2020
The conference on Neural Information Processing Systems (or NeurIPS, as it’s commonly known) is arguably the most important Machine Learning gathering of the year. Major research labs, like OpenAI and DeepMind, present some of their most notable work, and thousands of researchers have the opportunity to mingle and exchange ideas, alongside a range of winter sports activities.
Calling the papers that share the NeurIPS 2020 Best Paper Award the best research in AI for 2020 is as extreme as calling the LA Lakers the world champions. Another team in the world might be better, but they have the formalities done and dusted.
The thirty-fourth annual conference was held this past week as a pure virtual event. More than twelve thousand abstracts were submitted, which resulted in almost 9.500 full paper submissions. Although the number of submissions went up by 40% compared to 2019, only 20% of the papers were accepted (although someone could argue that this is a pretty good acceptance rate for NeurIPS).
This story summarizes the top three papers that shared the 2020 Best Paper Award. For those who could not attend the conference, there are several recorded presentations available online.