
An enhanced-state reinforcement learning algorithm that leverages user features, item features, and other valuable information to generate personalized fusion weights for each user-item pair, outperforming existing RL-based multi-task fusion methods in large-scale recommender systems.


coremsg

an-enhanced-state-reinforcement-learning-algorithm-for-optimizing-multi-task-fusion-in-large-scale-recommender-systems


An Enhanced-State Reinforcement Learning Algorithm for Optimizing Multi-Task Fusion in Large-Scale Recommender Systems
