OC-DeepIV combines interaction terms and orthogonal regularization, but its evaluation is circular and omits an outcome variable, so the reported causal effect estimates are not valid.
In the last 50 rounds, orthogonal constraint terms are gradually introduced to ensure stable optimisation and avoid excessive intervention in the learning process
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
econ.EM 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Orthogonality-Constrained Deep Instrumental Variable Model for Causal Effect Estimation
OC-DeepIV combines interaction terms and orthogonal regularization, but its evaluation is circular and omits an outcome variable, so the reported causal effect estimates are not valid.