Elastic-InfoGAN: Unsupervised Disentangled Representation Learning in Class-Imbalanced Data
arXiv:1910.01112
Abstract
We propose a novel unsupervised generative model that learns to disentangle object identity from other low-level aspects in class-imbalanced data. We first investigate the issues surrounding the assumptions about uniformity made by InfoGAN, and demonstrate its ineffectiveness to properly disentangle object identity in imbalanced data. Our key idea is to make the discovery of the discrete latent factor of variation invariant to identity-preserving transformations in real images, and use that as a signal to learn the appropriate latent distribution representing object identity. Experiments on both artificial (MNIST, 3D cars, 3D chairs, ShapeNet) and real-world (YouTube-Faces) imbalanced datasets demonstrate the effectiveness of our method in disentangling object identity as a latent factor of variation.
arXiv commentCamera ready version for NeurIPS 2020
References in corpus (18)
- SMOTE: Synthetic Minority Over-sampling Technique
- Auto-Encoding Variational Bayes
- A Simple Framework for Contrastive Learning of Visual Representations
- Categorical Reparameterization with Gumbel-Softmax
- A systematic study of the class imbalance problem in convolutional neural networks
- The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables
- Data-Efficient Image Recognition with Contrastive Predictive Coding
- Disentangling by Factorising
- Learning Discrete Representations via Information Maximizing Self-Augmented Training
- Disentangling factors of variation in deep representations using adversarial training
- Mixed Precision Quantization of ConvNets via Differentiable Neural Architecture Search
- MS-Celeb-1M: A Dataset and Benchmark for Large-Scale Face Recognition
- Joint Unsupervised Learning of Deep Representations and Image Clusters
- Class-Balanced Loss Based on Effective Number of Samples
- Deep Clustering via Joint Convolutional Autoencoder Embedding and Relative Entropy Minimization
- Deep Continuous Clustering
- Domain Adaptation for Semantic Segmentation via Class-Balanced Self-Training
- Semantic Bottleneck Scene Generation