no code implementations • 31 Oct 2018 • Boyu Zhang, Azadeh Davoodi, Yu-Hen Hu
Since the customized dataset is in general very small, the cost of training LE and GN would be much lower than that of re-training of GE.