Adversarial Constraint Learning for Structured Prediction
arXiv:1805.10561
Abstract
Constraint-based learning reduces the burden of collecting labels by having users specify general properties of structured outputs, such as constraints imposed by physical laws. We propose a novel framework for simultaneously learning these constraints and using them for supervision, bypassing the difficulty of using domain expertise to manually specify constraints. Learning requires a black-box simulator of structured outputs, which generates valid labels, but need not model their corresponding inputs or the input-label relationship. At training time, we constrain the model to produce outputs that cannot be distinguished from simulated labels by adversarial training. Providing our framework with a small number of labeled inputs gives rise to a new semi-supervised structured prediction model; we evaluate this model on multiple tasks --- tracking, pose estimation and time series prediction --- and find that it achieves high accuracy with only a small number of labeled inputs. In some cases, no labels are required at all.
arXiv commentTo appear at IJCAI 2018
References in corpus (11)
- Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
- Improved Training of Wasserstein GANs
- Improved Techniques for Training GANs
- Weakly- and Semi-Supervised Learning of a DCNN for Semantic Image Segmentation
- Data Programming: Creating Large Training Sets, Quickly
- Virtual Adversarial Training: A Regularization Method for Supervised and Semi-Supervised Learning
- Learning in Implicit Generative Models
- InfoGAIL: Interpretable Imitation Learning from Visual Demonstrations
- A Semantic Loss Function for Deep Learning with Symbolic Knowledge
- Pattern Decomposition with Complex Combinatorial Constraints: Application to Materials Discovery
- Max-Margin Deep Generative Models for (Semi-)Supervised Learning