Search
Duplicate

Off-Policy Evaluation

토픽 상세
Empty
구분
Empty
발표담당
오상곤, 조준기
발표자료
(Review) Modern_Recommendation_for_Advanced_Practitioners__Part1 (1).pdf
실습담당
Empty
주차
5주차(11/20)
랩업자료
Empty
자료2
Empty
목표
classic recommender system의 한계를 살펴보며 offline evaluation 컨셉이 어떻게 문제를 해결하는지 이해
Modern Recommendation for Advanced Practitioners
Part 1: Recommendation via Reward Modeling
Classic vs Modern: Recommendation as Autocomplete vs Recommendation as In tervention Policy
Recommendation Reward Modeling via Maximum Likelihood Models
Shortcomings of Classical value-based Recommendations