Augmented CycleGAN: Learning Many-to-Many Mappings from Unpaired Data
Authors: Amjad Almahairi, Sai Rajeshwar, Alessandro Sordoni, Philip Bachman, Aaron Courville
ICML 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We examine Augmented Cycle GAN qualitatively and quantitatively on several image datasets. |
| Researcher Affiliation | Collaboration | 1Montreal Institute for Learning Algorithms (MILA), Canada. 2Microsoft Research Montreal, Canada. 3CIFAR Fellow. Work partly done at MSR Montreal. |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Public code available at: https://github.com/aalmah/augmented_cyclegan |
| Open Datasets | Yes | Training data is composed of almost 50K shoe images with corresponding edges (Yu & Grauman, 2014; Zhu et al., 2016; Isola et al., 2017) |
| Dataset Splits | No | While it mentions a 'predefined test set of 200 samples', it does not provide comprehensive split information (e.g., percentages for training/validation/test or total counts for all splits) for all datasets used, nor does it refer to a standard split by citation for reproduction. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU models, CPU types, or memory specifications) used for running its experiments. |
| Software Dependencies | No | The paper does not specify particular software dependencies with version numbers (e.g., deep learning frameworks like TensorFlow or PyTorch, or Python versions). |
| Experiment Setup | No | The paper mentions hyperparameters like γ1 and γ2 are used to balance objectives, but their specific values are not provided. It refers to architectures like 'Res Net conditional image generators' and 'U-NET conditional image generators' but lacks detailed training settings such as learning rates, batch sizes, or number of epochs for the main model training. |