A comparison of Google's SlateQ algorithm with traditional Reinforcement Learning techniques for Recommendation Systems
In Google's recent SlateQ paper, they outline a tractable technique to applying Reinforcement Learning to Recommendation Systems. In this repository, we compare their SlateQ algorithm with other techniques via simulation using the recsim (GitHub repo) package.
For a high-level overview of the SlateQ algorithm, check out Craig Boutilier's talk at ICML 2019.
As governed by the recsim library, this repo uses Python 3.6.