Object-Oriented Dynamics Learning through Multi-Level Abstraction
arXiv:1904.07482
Abstract
Object-based approaches for learning action-conditioned dynamics has demonstrated promise for generalization and interpretability. However, existing approaches suffer from structural limitations and optimization difficulties for common environments with multiple dynamic objects. In this paper, we present a novel self-supervised learning framework, called Multi-level Abstraction Object-oriented Predictor (MAOP), which employs a three-level learning architecture that enables efficient object-based dynamics learning from raw visual observations. We also design a spatial-temporal relational reasoning mechanism for MAOP to support instance-level dynamics learning and handle partial observability. Our results show that MAOP significantly outperforms previous methods in terms of sample efficiency and generalization over novel environments for learning environment models. We also demonstrate that learned dynamics models enable efficient planning in unseen environments, comparable to true environment models. In addition, MAOP learns semantically and visually interpretable disentangled representations.
arXiv commentAccepted to the Thirthy-Fourth AAAI Conference On Artificial Intelligence (AAAI), 2020
References in corpus (10)
- Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
- Learning to Segment Object Candidates
- Interaction Networks for Learning about Objects, Relations and Physics
- Continuous Deep Q-Learning with Model-based Acceleration
- Relational Deep Reinforcement Learning
- Recurrent Environment Simulators
- Gotta Learn Fast: A New Benchmark for Generalization in RL
- Visual Interaction Networks
- Contingency-Aware Exploration in Reinforcement Learning
- Object-Oriented Dynamics Predictor