3 ms·

I sure agree there are many interesting things going on, there is no question about that. Also most of them are toy problems focused in some restricted domains,

I sure agree there are many interesting things going on, there is no question about that. Also most of them are toy problems focused in some restricted domains, while a huge bag of equally interesting real world problems is sitting untouched. And let me tell you, all those VC's that put in probably way north of $10B are not looking forward to more NIPS papers or yet another style transfer algorithm.

solomatov 8y ago
>Also most of them are toy problems focused in some restricted domains, while a huge bag of equally interesting real world problems is sitting untouched. It always starts with toy problems. Recognizing pictures from imagenet was also a toy problem back then.