Causal Overhypotheses

Causal overhypotheses represent abstract, higher-level beliefs about causal relationships that generalize across multiple situations. Current research focuses on developing computational models and algorithms, often within Bayesian frameworks, to learn and utilize these overhypotheses, particularly in active learning scenarios where agents strategically select actions to maximize long-term learning. This research aims to bridge the gap between human causal reasoning abilities and the capabilities of artificial intelligence, with implications for improving machine learning algorithms' ability to generalize and adapt to novel situations. The development of robust benchmarks and experimental paradigms, such as adaptations of the "blicket detector" task, is crucial for evaluating progress in this area.

Papers

December 11, 2023

Learning Unknown Intervention Targets in Structural Causal Models from Heterogeneous Data
Yuqin Yang, Saber Salehkaleybar, Negar Kiyavash
LeArning Abstract Causal Model Structural Causal Model Heterogeneous Data Latent Confounders Global Endogenous Variate Unknown Intervention Causal Overhypotheses

November 22, 2022

Causal Abstraction with Soft Interventions
Riccardo Massidda, Atticus Geiger, Thomas Icard, Davide Bacciu
Causal Model Metric Aware Abstraction Causal Abstraction Soft Intervention Causal Overhypotheses

June 20, 2022

Actively learning to learn causal relationships
Chentian Jiang, Christopher G. Lucas
Active Learning Causal Relationship Active Causal Causal Overhypotheses

June 16, 2022

Towards Understanding How Machines Can Learn Causal Overhypotheses
Eliza Kosoy, David M. Chan, Adrian Liu, Jasmine Collins, Bryanna Kaufmann, Sandy Han Huang, Jessica B. Hamrick, John Canny, Nan Rosemary Ke, Alison Gopnik
New Machine Causal Inference Causal Relationship Causal Overhypotheses

February 21, 2022

Learning Causal Overhypotheses through Exploration in Children and Computational Models
Eliza Kosoy, Adrian Liu, Jasmine Collins, David M Chan, Jessica B Hamrick, Nan Rosemary Ke, Sandy H Huang, Bryanna Kaufmann, John Canny, Alison Gopnik
Reinforcement Learning Environment Exploration Causal Structure Nine Year Old Child Computational Model Causal Learning Optimal Reinforcement Learning Causal Overhypotheses

Causal Overhypotheses

Papers

Learning Unknown Intervention Targets in Structural Causal Models from Heterogeneous Data

Causal Abstraction with Soft Interventions

Actively learning to learn causal relationships

Towards Understanding How Machines Can Learn Causal Overhypotheses

Learning Causal Overhypotheses through Exploration in Children and Computational Models