Neuroevolutionary Feature Representations for Causal Inference
View / Open Files
Publication Date
2022Journal Title
Computational Science - ICCS 2022: 22nd International Conference,
London, United Kingdom, June 21-23, 2022, Proceedings, Part II, pp. 3-10
Conference Name
International Conference on Computational Science
ISSN
0302-9743
ISBN
9783031087530
Publisher
Springer International Publishing
Type
Conference Object
This Version
AM
Metadata
Show full item recordCitation
Burkhart, M., & Ruiz, G. (2022). Neuroevolutionary Feature Representations for Causal Inference. Computational Science - ICCS 2022: 22nd International Conference,
London, United Kingdom, June 21-23, 2022, Proceedings, Part II, pp. 3-10 https://doi.org/10.1007/978-3-031-08754-7_1
Abstract
Within the field of causal inference, we consider the problem of estimating
heterogeneous treatment effects from data. We propose and validate a novel
approach for learning feature representations to aid the estimation of the
conditional average treatment effect or CATE. Our method focuses on an
intermediate layer in a neural network trained to predict the outcome from the
features. In contrast to previous approaches that encourage the distribution of
representations to be treatment-invariant, we leverage a genetic algorithm that
optimizes over representations useful for predicting the outcome to select
those less useful for predicting the treatment. This allows us to retain
information within the features useful for predicting outcome even if that
information may be related to treatment assignment. We validate our method on
synthetic examples and illustrate its use on a real life dataset.
Sponsorship
Adobe Inc. (San José, Calif., USA)
Embargo Lift Date
2023-06-15
Identifiers
External DOI: https://doi.org/10.1007/978-3-031-08754-7_1
This record's URL: https://www.repository.cam.ac.uk/handle/1810/337412
Statistics
Total file downloads (since January 2020). For more information on metrics see the
IRUS guide.