Generative Well-intentioned Networks

Authors: Justin Cosentino, Jun Zhu

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

Reproducibility Variable Result LLM Response
Research Type Experimental We then empirically evaluate the effectiveness of the proposed framework in Section 5. Lastly, we discuss related works in Section 6.We evaluate the WGWIN-GP using the training procedure outlined in Section 4 and the inference method illustrated in Figure 1.
Researcher Affiliation Academia Justin Cosentino, Jun Zhu Dept. of Comp. Sci. & Tech., Institute for AI, THBI Lab, BNRist Center, State Key Lab for Intell. Tech. & Sys., Tsinghua University, Beijing, China
Pseudocode Yes Algorithm 1: WGWIN with gradient and transformation penalty.
Open Source Code No The paper does not provide an explicit statement about releasing code or a link to a code repository for the described methodology.
Open Datasets Yes We use two different datasets in our experiments: the MNIST handwritten digits [23] dataset and the Fashion-MNIST clothing dataset [41].
Dataset Splits Yes We further split both training sets into a 50,000 image training set and 10,000 image validation set.
Hardware Specification Yes We trained and evaluated the models using NVIDIA Ge Force GTX TITAN X GPUs.
Software Dependencies No The network is implemented using Tensor Flow Probability [7]. No specific version numbers for software dependencies are provided.
Experiment Setup Yes The BNN trained for 30 epochs using a learning rate of 0.001 and batch size of 128. The GWIN trained for 200,000 iterations using the default hyperparameters listed in Algorithm 1. Both the generator and critic used a learning rate of 0.0001 and batch size of 128.