Abstract:
This paper presents a comprehensive platform named AutoRec, which can help developers build effective and explainable recommender models all in one platform. It implements several well-known and state-of-art deep learning models in item recommendation scenarios, a AutoML framework with a package of search algorithms for automatically tuning of hyperparameters, and several instance-level interpretation methods to enable the explainable recommendation. The main advantage of AutoRec is the integration of AutoML and explainable AI abilities into the deep learning based recommender algorithms platform.