Deep Generative Models with Learnable Knowledge Constraints
arXiv:1806.09764
Abstract
The broad set of deep generative models (DGMs) has achieved remarkable advances. However, it is often difficult to incorporate rich structured domain knowledge with the end-to-end DGMs. Posterior regularization (PR) offers a principled framework to impose structured constraints on probabilistic models, but has limited applicability to the diverse DGMs that can lack a Bayesian formulation or even explicit density evaluation. PR also requires constraints to be fully specified a priori, which is impractical or suboptimal for complex knowledge with learnable uncertain parts. In this paper, we establish mathematical correspondence between PR and reinforcement learning (RL), and, based on the connection, expand PR to learn constraints as the extrinsic reward in RL. The resulting algorithm is model-agnostic to apply to any DGMs, and is flexible to adapt arbitrary constraints with the model jointly. Experiments on human image generation and templated sentence generation show models with learned knowledge constraints by our algorithm greatly improve over base generative models.
arXiv commentNeural Information Processing Systems (NeurIPS) 2018
Cited by in corpus (16)
- A Survey of Knowledge-Enhanced Text Generation
- Constrained Deep Networks: Lagrangian Optimization via Log-Barrier Extensions
- Controllable Data Generation by Deep Learning: A Review
- Texar: A Modularized, Versatile, and Extensible Toolkit for Text Generation
- Efficient Generation of Structured Objects with Constrained Adversarial Networks
- Deep Reasoning Networks: Thinking Fast and Slow
- Improving GAN Training with Probability Ratio Clipping and Sample Reweighting
- Unsupervised Editing for Counterfactual Stories
- fAshIon after fashion: A Report of AI in Fashion
- A Roadmap to Domain Knowledge Integration in Machine Learning
- KG-GAN: Knowledge-Guided Generative Adversarial Networks
- Knowledge-driven Encode, Retrieve, Paraphrase for Medical Image Report Generation
- Kernel-Guided Training of Implicit Generative Models with Stability Guarantees
- Training Deep Normalizing Flow Models in Highly Incomplete Data Scenarios with Prior Regularization
- Distantly Supervised Semantic Text Detection and Recognition for Broadcast Sports Videos Understanding
- Domain Constraint Approximation based Semi Supervision