learning point processes by means of optimal transport and wasserstein distance "Wasserstein Learning of Deep Generative Point Process Models" ############################################
Code accompanying the paper "Wasserstein Learning of Deep Generative Point Process Models"
If the code helps your research, please cite our work.
@inproceedings{xiao2017wasserstein,
title={Wasserstein Learning of Deep Generative Point Process Models.},
author={Xiao, Shuai and Farajtabar, Mehrdad and Ye, Xiaojing and Yan, Junchi and Yang, Xiaokang and Song, Le and Zha, Hongyuan},
booktitle={Advances in Neural Information Processing Systems 30},
year={2017}
}
- Computer with Linux or OSX
- Language: TensorFlow 1.0
- GPU is strongly recommended when training.
- python ppwgan DATA NUM_Seq Lambda # for CPU
- CUDA_VISIBLE_DEVICES=0 python ppwgan DATA NUM_Seq Lambda # for GPU
- For bugs and questions, contact: benjaminforever at sjtu.edu.cn
Point processes are becoming very popular in modeling asynchronous sequential data due to their sound mathematical foundation and strength in modeling a variety of real-world phenomena. Currently, they are often characterized via intensity function which limits model’s expressiveness due to unrealistic assumptions on its parametric form used in practice. Furthermore, they are learned via maximum likelihood approach which is prone to failure in multi-modal distributions of sequences. In this paper, we propose an intensity-free approach for point processes modeling that transforms nuisance processes to a target one. Furthermore, we train the model using a likelihood-free leveraging Wasserstein distance between point processes. Experiments on various synthetic and real-world data substantiate the superiority of the proposed point process model over conventional ones.