Would you mind linking to the 21 Jump Street scene? I'm very curious. As an American nerd in my late 30s, I definitely recall being relentlessly teased as a kid. I don't have kids of my own now and so I don't really know how and if things have changed, but if true this would be a very positive development!
"E-scooters aren't a reliable way to get anywhere yet, and who knows if they'll ever be, not to mention that they are not for everyone. My grandmother is not going to ride one -- nor my wife, for that matter, nor should the kids. But the Subway is a common denominator."
I used to think so, but some European cities really do offer counterexamples. I'm thinking of places like Munich, Vienna, and Copenhagen. It's not uncommon to see people there who, by American stereotypes, wouldn't be expected to ride scooters: moms with kids, men in suits, etc. Perhaps the urban cultural gap is so vast that what you're saying is indeed true of the US, but I wouldn't take it as a given.
It looks that way because they're moving rapidly from one face configuration to another. But there's no way that's happening by random. I would guess that even just holding the cube constant in a dynamic grip is quite difficult.
I would say the biggest thing is obviously the architecture, coupling LSTMs with the geometric units that spit out the actual 3D structure that can then be directly optimized via the dRMSD loss function. That's the biggest point of distinction from everything else out there (no contact map prediction, etc.) So it really is about end-to-end differentiability IMO, which hasn't been done before.
As for why it took so long, it is and it is not fine-tuning. Getting RGNs to train _at all_ was a rather difficult process, and required a lot of finicking around. But since I got them working, I haven't actually spent all that much time fine-tuning them, and so I expect there to be a lot of low-hanging fruit in terms of optimizing performance (starting from the baseline I found.)
Re drug discovery, often times in “rational” drug design, medicinal chemists try to make small molecules that bind snuggly into a binding pocket on the protein. Having the structure of the protein aids greatly in that process.
I do think however that protein folding is very much understudied in the ML community, relative to say the big three of vision, NLP, and speech. The lack of standardized data sets and benchmarks, not to mention the need for domain knowledge, have made it difficult to get into the field
Hi! I’m the author of the paper. Not sure why you say Rosetta isn’t mentioned? It’s extensively referenced throughout the paper, discussed in the discussion section, and is one of the top 5 CASP servers compared to in the results section.
Also as for how it’s different from what’s described in the paper, that’s the topic of the introduction of the paper. Rosetta uses both fragment assembly and co-evolution methods.
I took CS221 from Andrew in 2006 (or was it 2007?) Even more has changed since then ;-) It was my second ML course, after taking Daphne Koller's punishing CS229. Right then though I knew ML will sweep the world pretty soon.