Isolating Sources of Disentanglement in Variational Autoencoders
Authors: Ricky T. Q. Chen, Xuechen Li, Roger B. Grosse, David K. Duvenaud
NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We perform extensive quantitative and qualitative experiments, in both restricted and non-restricted settings, and show a strong relation between total correlation and disentanglement, when the model is trained using our framework. |
| Researcher Affiliation | Academia | Ricky T. Q. Chen, Xuechen Li, Roger Grosse, David Duvenaud University of Toronto, Vector Institute |
| Pseudocode | No | The paper does not contain any clearly labeled pseudocode or algorithm blocks. |
| Open Source Code | Yes | Code is available at . |
| Open Datasets | Yes | We perform quantitative evaluations with two datasets, a dataset of 2D shapes [39] and a dataset of synthetic 3D faces [40]... [39] refers to: dsprites: Disentanglement testing sprites dataset. https://github.com/deepmind/dsprites-dataset/, 2017. |
| Dataset Splits | No | The paper mentions using datasets for experiments but does not provide specific details on how these datasets were split into training, validation, or test sets (e.g., percentages or sample counts). |
| Hardware Specification | No | The paper does not provide any specific details about the hardware (e.g., GPU models, CPU types, memory) used to conduct the experiments. |
| Software Dependencies | No | The paper does not list specific software dependencies with version numbers (e.g., Python 3.x, TensorFlow 1.x, PyTorch 1.x) that were used for the experiments. |
| Experiment Setup | Yes | We used β = 4 for β-VAE and β = 6 for β-TCVAE, based on modes in Figure 2. For Info GAN, we used 5 continuous latent codes and 5 noise variables. Other settings are chosen following those suggested by [6], but we also added instance noise [41] to stabilize training. ... we tuned β [1, 80] and used double the number of iterations for Factor VAE. Note that while β-VAE, Factor VAE and β-TCVAE use a fully connected architecture for the d Sprites dataset, Info GAN uses a convolutional architecture for increased stability. |