Structured Generative Adversarial Networks
Authors: Zhijie Deng, Hao Zhang, Xiaodan Liang, Luona Yang, Shizhen Xu, Jun Zhu, Eric P. Xing
NeurIPS 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We assess SGAN by evaluating its trained networks, and its performance on downstream tasks. We show that SGAN delivers a highly controllable generator, and disentangled representations; it also establishes start-of-the-art results across multiple datasets when applied for semi-supervised image classification (1.27%, 5.73%, 17.26% error rates on MNIST, SVHN and CIFAR-10 using 50, 1000 and 4000 labels, respectively). To empirically evaluate SGAN, we first define a mutual predictability (MP) measure to evaluate the disentanglability of various DGMs, and show that in terms of MP, SGAN outperforms all existing models that are able to infer the latent code z across multiple image datasets. In the semisupervised image classification task, SGAN outperforms strong baselines, and establishes new state-of-the-art results on MNIST, SVHN and CIFAR-10 dataset. To empirically validate this, we deploy SGAN for semi-supervised classification on MNIST, SVHN and CIFAR-10, and compare the test errors of C to strong baselines in Table 2. To figure out how each module in SGAN contributes to the final results, we conduct an ablation study in Fig.3. |
| Researcher Affiliation | Collaboration | 1Tsinghua University, 2Carnegie Mellon University, 3Petuum Inc. |
| Pseudocode | Yes | Algorithm 1 Training Structured Generative Adversarial Networks (SGAN). |
| Open Source Code | Yes | The code is publicly available at https://github.com/thudzj/Structured GAN. |
| Open Datasets | Yes | We evaluate SGAN on three image datasets: (1) MNIST [14]: we use the 60K training images as unlabeled data, and sample n {20, 50, 100} labels for semi-supervised learning following [12, 27], and evaluate on the 10K test images. (2) SVHN [20]: a standard train/test split is provided, where we sample n = 1000 labels from the training set for semi-supervised learning [27, 15, 5]. (3) CIFAR-10: a challenging dataset for conditional image generation that consists of 50K training and 10K test images from 10 object classes. We randomly sample n = 4000 labels [27, 28, 15] for semi-supervised learning. |
| Dataset Splits | No | The paper describes training and test sets but does not explicitly provide details for a separate validation dataset split (e.g., percentages or sample counts for validation). |
| Hardware Specification | No | The paper mentions "distributed acceleration provided by Poseidon [33] which parallelizes line 7-8 and 10-12 of Algorithm. 1." and "efficient GPU-based deep learning on multiple machines". However, it does not specify exact GPU models, CPU models, or other specific hardware configurations. |
| Software Dependencies | No | We implement SGAN using Tensor Flow [1] and Theano [2] with distributed acceleration provided by Poseidon [33]... The neural network architectures of C, G and Dxy mostly follow those used in Triple GAN [15] and we design I and Dxz according to [5]... The paper mentions software like TensorFlow and Theano but does not provide specific version numbers for them or any other libraries/solvers. |
| Experiment Setup | Yes | We adopt the same neural network architectures and hyper-parameter settings from [15]... The neural network architectures of C, G and Dxy mostly follow those used in Triple GAN [15] and we design I and Dxz according to [5] but with shallower structures to alleviate the training costs. Properly weighting the losses of the four games in SGAN during training may lead to performance improvement. However, we simply set them equal without heavy tuning. |