Reviews: Deep Learning for Predicting Human Strategic Behavior

The problem is well-motivated and important. The proposed architecture is novel and appropriate for these type of problems and could lead to better prediction of strategic interaction. It is clearly beating the state of the art; though it is not entirely clear how/why. The paper is written well (except for some ambiguities that could be addressed in the response and camera-ready), which helps its potential impact. There are a number of seemingly arbitrary choices made throughout the paper.