Weakly-Supervised Disentanglement Without Compromises
Authors: Francesco Locatello, Ben Poole, Gunnar Raetsch, Bernhard Schölkopf, Olivier Bachem, Michael Tschannen
ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | First, we theoretically show that only knowing how many factors have changed, but not which ones, is sufficient to learn disentangled representations. Second, we provide practical algorithms that learn disentangled representations from pairs of images without requiring annotation of groups, individual factors, or the number of factors that have changed. Third, we perform a large-scale empirical study and show that such pairs of observations are sufficient to reliably learn disentangled representations on several benchmark data sets. |
| Researcher Affiliation | Collaboration | 1Department of Computer Science, ETH Zurich 2Max Planck Institute for Intelligent Systems 3Google Research, Brain Team. Correspondence to: <francesco.locatello@inf.ethz.ch>. |
| Pseudocode | No | The paper describes models and processes mathematically and in prose, but does not include any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not contain any explicit statements about releasing source code for the described methodology, nor does it provide a link to a code repository. |
| Open Datasets | Yes | We use the five data sets where the observations are generated as deterministic functions of the factors of variation: d Sprites (Higgins et al., 2017a), Cars3D (Reed et al., 2015), Small NORB (Le Cun et al., 2004), Shapes3D (Kim & Mnih, 2018), and the real-world robotics data set MPI3D (Gondal et al., 2019). |
| Dataset Splits | No | The paper mentions training and test sets but does not provide specific percentages, sample counts, or a clear methodology for validation splits for the main model training, only for downstream tasks with varying sample sizes (10/100/1000/10 000). |
| Hardware Specification | Yes | Our experiments required 5.85 GPU years (NVIDIA P100). |
| Software Dependencies | No | The paper mentions various models (e.g., β-VAE, Factor VAE, β-TCVAE) and classifiers (logistic regression, GBT), but does not provide specific version numbers for any software libraries, frameworks, or languages used. |
| Experiment Setup | Yes | For each approach we consider six values for the regularization strength and 10 random seeds, training a total of 6000 weakly-supervised models. We perform model selection using the weakly-supervised reconstruction loss (i.e., the sum of the first two terms in (6)). |