Implementation of DP-Spectral. DP-Spectral is a spectral-based network embedding algorithm that employs "degree-punishment principle" for improving the scale-free property preservation in the embedded space.
For detailed descriptions of the model, please refer to
Rui Feng, Yang Yang, Wenjie Hu, Fei Wu, and Yueting Zhuang. Representation Learning for Scale-free Networks. In Proceedings of the 32nd AAAI Conference on Artificial Intelligence (AAAI'18), 2018, pages 282-289.