1 paper
Yoav Kurtz, Noga Bar, Raja Giryes
As neural networks become deeper, the redundancy within their parameters increases. This phenomenon has led to several methods that attempt to reduce the correlation between convol…