142 citations · 512 across the 30 of their papers we have counts for
1 paper · 1 filter
Armen Aghajanyan
The downfall of many supervised learning algorithms, such as neural networks, is the inherent need for a large amount of training data. Although there is a lot of buzz about big da…