Entry
Breaking Feedback Loops in Recommender Systems with Causal Inference
Karl Krauth, Yixin Wang, Michael I. Jordan
CAFL algorithm provably breaks feedback loops via causal inference; applies to any recommender that optimizes a training loss.
·causal inference ·CAFL ·feedback loops ·intervention distribution