ImageNet

“I still think that ImageNet was more impactful in the vision community, and the reason for that is a few. The first is that ImageNet also allowed for reproducible evaluation, right? You know, OpenX as an effort was more about making data available for kind of …”

“Distillation was originally motivated because we were seeing that we had a very large image data set at the time, you know, three hundred million images that we could train on with, you know, I forget, like 20,000 categories or something, so much bigger than I…”

“We curated very carefully, fifteen million images on the internet, created a taxonomy of 22,000 concepts, borrowing other researchers' work, like linguists' work on WordNet, and it's a particular way of dictionary words.”

“Along comes AlexNet. Its error rate was 15%. Still high, but a 10% leap from the previous best being a 25% error rate all the way down to 15 in one year. A leap like that had never happened before.”

“One piece of intuition around that is that you can almost think about structure prediction as The ImageNet moment for the field, where with structure prediction, we are asking a model to go from sequence to a predicted structure, and it's sort of like a classi…”

“And yeah, I was one of the coauthors of ImageNet too.”

“The 2012 AlexNet paper on ImageNet Challenge is literally a very classic model, and that is the convolutional neural network model, and that was published in 19 eighties, the first paper. I remember as a graduate student learning that, and it more or less also…”

“The best way to energize in our world is open source. You make this invaluable resource available and invite everybody to participate in a benchmark test.”

“I think the error rate went from mislabeling images 25% of the time to suddenly only mislabeling them 15% of the time, and that was like a huge leap over the tiny incremental progress that had been made along the way.”

“And we're now kind of going through this again, I think, as you look at things like Dalai, or Dali, or however you pronounce it, and Imogen, and GPT-III, of again, here's something that looks like a breakthrough. Here are all these amazing demos, and the demos…”

“Like the, if you look at the results, like image net's a good, it's a good example where the, you know, it's been the error rates were 30% or something, and now are better than humans.”

“We estimate humans can do this with about five percent error, so we're already better at identifying things and objects than humans.”

“That's the essential data set that made deep learning happen.”

“A team from University of Toronto, led by Geoffrey Hinton, and that's, like, the team was Alex Krzyzewski and Ilya Suskaver. They actually got To something like 15%. So let's say all other teams, they were like at 25%, the difference was one percent. Yeah. And…”

“Within several years, people got down, I believe, to three percent error, and that's essentially superhuman performance.”

“There were, was no training data in the ImageNet data set of the New York City subway system.”