ROBUST ESTIMATION VIA GENERATIVE ADVERSARIAL NETWORKS

Authors: Chao GAO, jiyi LIU, Yuan YAO, Weizhi ZHU

ICLR 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Numerical experiments in Section 5 are provided to show the success of these GANs.
Researcher Affiliation Academia Chao Gao Department of Statistics University of Chicago Chicago, IL 60637 USA chaogao@galton.uchicago.edu; Jiyi Liu Department of Statistics and Data Science Yale University New Haven, CT 06511 USA jiyi.liu@yale.edu; Yuan Yao & Weizhi Zhu Department of Mathematics Hong Kong University of Science and Technology Kowloon, Hong Kong yuany@ust.hk; wzhuai@connect.ust.hk
Pseudocode Yes Algorithm 1 JS-GAN: argminη maxw[ 1/n Pn i=1 log Dw(Xi) + E log(1 Dw(Gη(Z)))]
Open Source Code Yes A PyTorch implementation is available at https://github.com/zhuwzh/Robust-GAN-Center.
Open Datasets No The paper utilizes synthetically generated data (e.g., from (1 ϵ)N(0p, Ip) + ϵQ) and describes the data generation process rather than using or providing a publicly available dataset.
Dataset Splits No The paper primarily uses synthetically generated data and does not specify traditional train/validation/test splits for publicly available datasets.
Hardware Specification No The paper does not specify the hardware used for running the experiments.
Software Dependencies No The paper mentions 'PyTorch implementation' but does not specify its version or any other software dependencies with version numbers.
Experiment Setup Yes Table 5: Choices of hyper-parameters. The parameter λ is the penalty factor for the regularization term (17) and other parameters are listed in Algorithm 1. We apply Xavier initialization (Glorot & Bengio, 2010) for both JS-GAN and TV-GAN trainings.