Skip to yearly menu bar Skip to main content


Poster

Deep Generative Models with Learnable Knowledge Constraints

Zhiting Hu · Zichao Yang · Russ Salakhutdinov · LIANHUI Qin · Xiaodan Liang · Haoye Dong · Eric Xing

Room 210 #32

Keywords: [ Generative Models ]


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 {\it 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.

Live content is unavailable. Log in and register to view live content