Deep Learning doesn't necessarily require a huge amount of data. References? I mean, I know "one shot" and similar approaches but as far as I know, these invol

Deep Learning doesn't necessarily require a huge amount of data. References? I mean, I know "one shot" and similar approaches but as far as I know, these involve extending a neural network that has been already trained, on massive data, to a little bit more.