Demo-Stochastic-VS-Deterministic

Unofficial implementation of the demo in "Reasons for the Superiority of Stochastic Estimators over Deterministic Ones: Robustness, Consistency and Perceptual Quality".

@article{sto_vs_det,
  title={Reasons for the Superiority of Stochastic Estimators over Deterministic Ones: Robustness, Consistency and Perceptual Quality},
  author={Ohayon, Guy and Adrai, Theo and Elad, Michael and Michaeli, Tomer},
  journal={arXiv preprint arXiv:2211.08944},
  year={2022}
}

Training, Inference, & Visualization

• Training: python train_gan.py --stochastic --robust
• Inference: python infer_gan.py --stochastic --robust
• Visualization: python vis_pred.py --stochastic --robust -n #NUM_POINTS

Note: --stochastic --robust are optional. --stochastic means using stochastic algorithm, and --robust means using robust loss.

Acknowledgement

We use some codes from PyTorch-GAN.