ALICE: Towards Understanding Adversarial Learning for Joint Distribution Matching
Authors: Chunyuan Li, Hao Liu, Changyou Chen, Yuchen Pu, Liqun Chen, Ricardo Henao, Lawrence Carin
NeurIPS 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on synthetic and real data demonstrate that ALICE is significantly more stable to train than ALI, in that it consistently yields more viable solutions (good generation and good reconstruction), without being too sensitive to perturbations of the model architecture, i.e., hyperparameters. |
| Researcher Affiliation | Academia | Chunyuan Li1, Hao Liu2, Changyou Chen3, Yunchen Pu1, Liqun Chen1, Ricardo Henao1 and Lawrence Carin1 1Duke University 2Nanjing University 3University at Buffalo cl319@duke.edu |
| Pseudocode | No | The paper describes methods textually and mathematically but does not include any explicit pseudocode or algorithm blocks. |
| Open Source Code | Yes | The code to reproduce these experiments is at https://github.com/Chunyuan LI/ALICE |
| Open Datasets | Yes | Two image-to-image translation tasks are considered. (i) Car-to-Car [24]: each domain (x and z) includes car images in 11 different angles... (ii) Edge-to-Shoe [25]: x domain consists of shoe photos and z domain consists of edge images... This is demonstrated on the Celeb A face dataset [27]. |
| Dataset Splits | No | The paper explicitly states "We train on 2048 samples, and test on 1024 samples." for the toy dataset. While a grid search for hyperparameters is mentioned, indicating some form of validation, an explicit "validation set" or clear train/validation/test splits are not provided for all datasets, nor are the methods for defining splits described in detail. |
| Hardware Specification | No | The paper does not provide specific details regarding the hardware used for running the experiments (e.g., GPU/CPU models, memory specifications). |
| Software Dependencies | No | The paper does not list specific software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow versions). |
| Experiment Setup | No | While the paper mentions using a "grid search over a set of architectural choices and hyper-parameters" and "equal weighting" for some terms, it does not explicitly provide concrete hyperparameter values (e.g., learning rate, batch size, specific network architectures) or detailed training configurations in the main text. |