
Episode 13: Jonathan Frankle, MIT, on the lottery ticket hypothesis and the science of deep learning
Generally Intelligent
00:00
I Think We Need More Than That in the Field of Science
Science is an emerging topic, and reminding the reviewers that this is important work. I'm working with a couple of colleagues right now to just think more broadly about how we should categorize papers. The first is an engineering paper, where you propose a technique that makes something better and works well. There are science papers where you ask a question, and really the evaluation of the paper should look something like, what is your hypothesis? What is the background for why you came ovr this hypothesis? That should be the criteria on which we determine whether we accept.
Transcript
Play full episode