Causal Induction from Visual Observations for Goal Directed Tasks

Abstract

Causal reasoning has been an indispensable capability for humans and other intelligent animals to interact with the physical world. In this work, we propose to endow an artificial agent with the capability of causal reasoning for completing goal-directed tasks. We develop learning-based approaches to inducing causal knowledge in the form of directed acyclic graphs, which can be used to contextualize a learned goal-conditional policy to perform tasks in novel environments with latent causal structures. We leverage attention mechanisms in our causal induction model and goal-conditional policy, enabling us to incrementally generate the causal graph from the agent's visual observations and to selectively use the induced graph for determining actions. Our experiments show that our method effectively generalizes towards completing new tasks in novel environments with previously unseen causal structures.

13 pages, 6 figures

References in corpus (11)

• Hindsight Experience Replay
• A Meta-Transfer Objective for Learning to Disentangle Causal Mechanisms
• Meta Reinforcement Learning with Latent Variable Gaussian Processes
• Learning Plannable Representations with Causal InfoGAN
• Causal Reasoning from Meta-reinforcement Learning
• CausalGAN: Learning Causal Implicit Generative Models with Adversarial Training
• Visual Causal Feature Learning
• Environment Probing Interaction Policies
• Woulda, Coulda, Shoulda: Counterfactually-Guided Policy Search
• Structural Agnostic Modeling: Adversarial Learning of Causal Graphs
• Gradient-Based Neural DAG Learning

Cited by in corpus (17)

• Unbiased Scene Graph Generation from Biased Training
• iPerceive: Applying Common-Sense Reasoning to Multi-Modal Dense Video Captioning and Video Question Answering
• Invariant Policy Optimization: Towards Stronger Generalization in Reinforcement Learning
• Who Make Drivers Stop? Towards Driver-centric Risk Assessment: Risk Object Identification via Causal Inference
• Disentanglement via Mechanism Sparsity Regularization: A New Principle for Nonlinear ICA
• Human Trajectory Prediction via Counterfactual Analysis
• Systematic Evaluation of Causal Discovery in Visual Model Based Reinforcement Learning
• Amortized learning of neural causal representations
• Generative Interventions for Causal Learning
• iReason: Multimodal Commonsense Reasoning using Videos and Natural Language with Interpretability
• Disentangling causal effects for hierarchical reinforcement learning
• Visual Superordinate Abstraction for Robust Concept Learning
• Causal Structure Learning: a Bayesian approach based on random graphs
• Towards intervention-centric causal reasoning in learning agents
• A Unified View of Algorithms for Path Planning Using Probabilistic Inference on Factor Graphs
• Using Human-Guided Causal Knowledge for More Generalized Robot Task Planning
• Did I do that? Blame as a means to identify controlled effects in reinforcement learning