13.01.2026
New paper on transfer learning of individual treatment effects
We submitted a paper showing that individual treatment effect estimation from the TARNet model can be improved in small or systematically different target datasets by transferring causal knowledge from larger source data. Across simulations and empirical data, we show that transfer learning reduces ITE estimation error and attenuates bias particularly when a large, unbiased source dataset is available and the target sample is small or potentially biased.
Aydin, S. B., & Brandt, H. (submitted). Advantages and limitations in the use of transfer learning for individual treatment effects in causal machine learning. Article Github
Share
