Reviews: Sample Efficient Active Learning of Causal Trees

The authors proposed a suite of algorithms for learning the structure of the causal graph under different assumptions (infinite and finite interventional sample, single vs. K intervention, non-manipulable variables). The assumption about the type of underlying causal graphs is quite stringent: a tree with no v-structure. Authors do not provide a compelling real-world example where this assumption makes sense. Nevertheless, this work seems to provide a theoretical insight to the very specific class of problems. Overall the paper is written clearly for readers to follow without any interruptions in general (there are some issues with how the paper is organized and I will talk about this below.)