Qinglin Jia

Existing deep feature selection methods for recommender systems suffer from layer, baseline, and approximation biases, leading to suboptimal feature selection, which FairFS effectively mitigates.