Contrastive Learning of Structured World Models

2019/11/27 by Thomas Kipf, Kipf, Thomas, Elise van der Pol +3 · 7 citations
Computer Science · #Advanced Image and Video Retrieval Techniques #Artificial Intelligence (cs.AI) #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Multimodal Machine Learning Applications

paper · pdf · doi:10.48550/arxiv.1911.12247

openalex publication_date 2019/11/27 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

A structured understanding of our world in terms of objects, relations, and hierarchies is an important component of human cognition. Learning such a structured world model from raw sensory data remains a challenge. As a step towards this goal, we introduce Contrastively-trained Structured World Models (C-SWMs). C-SWMs utilize a contrastive approach for representation learning in environments with compositional structure. We structure each state embedding as a set of object representations and their relations, modeled by a graph neural network. This allows objects to be discovered from raw pixel observations without direct supervision as part of the learning process. We evaluate C-SWMs on compositional environments involving multiple interacting objects that can be manipulated independently by an agent, simple Atari games, and a multi-object physics simulation. Our experiments demonstrate that C-SWMs can overcome limitations of models based on pixel reconstruction and outperform typical representatives of this model class in highly structured environments, while learning interpretable object-based representations.

Citations

• Adam: A Method for Stochastic Optimization
• Matplotlib: A 2D Graphics Environment
• Gated Graph Sequence Neural Networks
• Semi-Supervised Classification with Graph Convolutional Networks
• Graph networks as learnable physics engines for inference and control
• Learning Latent Dynamics for Planning from Pixels
• Model-Based Reinforcement Learning for Atari
• Data-Efficient Image Recognition with Contrastive Predictive Coding
• COBRA: Data-Efficient Model-Based RL through Unsupervised Object Discovery and Curiosity-Driven Exploration

Cited by

• ORGAN: Object-Centric Representation Learning Using Cycle Consistent Generative Adversarial Networks
• Soft Geometric Inductive Bias for Object Centric Dynamics
• PIPHEN: Physical Interaction Prediction with Hamiltonian Energy Networks
• Multi-step Predictive Coding Leads To Simplicity Bias
• MACIE: Multi-Agent Causal Intelligence Explainer for Collective Behavior Understanding
• Learning Interactive World Model for Object-Centric Reinforcement Learning
• From Observation to Insight: Mechanistic World Models and the Quest for Autonomous Discovery

Related

• Learning Structured Representations of Visual Scenes
• Generalization properties of contrastive world models
• ReCoRe: Regularized Contrastive Representation Learning of World Model
• Robust and Controllable Object-Centric Learning through Energy-based Models
• Dyn-O: Building Structured World Models with Object-Centric Representations