Meta-Learned User Preference Estimator with Attention Network for Cold-Start Recommendation

Abstract One crucial challenge in the recommendation research field is cold-start problem. Meta-learning a feasible algorithm to reduce error of because it can adjust new tasks rapidly through relatively few updates. However, meta-learning does not take diverse interests users into account, which limits performance improvement scenarios. We proposed model called attentional meta-learned user preference estimator that combines attention network and meta-learning. This method enhances ability modelling personalized interest by learning weights between items based on mechanism, then improves recommendation. validated with two publicly available datasets field. Compared three benchmark methods, reduces mean absolute at least 2.3% root square 2.5%.