The paper describing the Vaegan approach will appear at GLSVLSI 2024.
This repository contains the code and supporting documents for the VAEGAN.
-
Metrics: Includes the evaluation metrics such as Maximum Mean Discrepancy (MMD), Sum of Squared Differences (SSD), Precision-Recall Density (PRD), and Cosine Similarity (COSS) for both MLPVAE and DCGAN models.
-
CustomDataset: Manages the creation of batches of files for the networks, facilitating efficient data handling.
-
DataTransformation: Transforms float and integer numbers into a 32-bit fixed-point binary number, with 12 bits allocated for the fractional part, optimizing data representation for computational processes.
Details on how to use these modules and scripts will be provided in subsequent sections or documents within this repository.
For more information or queries, please contact Yuchao Liao at [email protected].
If you use the Vaegan in your work, please cite it as follows:
@article{liao2024skip,
title={Skip the Benchmark: Generating System-Level High-Level Synthesis Data using Generative Machine Learning},
author={Liao, Yuchao and Adegbija, Tosiron and Lysecky, Roman and Tandon, Ravi},
journal={arXiv preprint arXiv:2404.14754},
year={2024}
}