Artwork Personalization at Netflix Fernando Amat RecSys2018
10 likes•3,959 views
The document discusses Netflix's use of personalized artwork optimization to enhance content discovery for its 130 million global members. It describes the algorithms employed to recommend suitable images based on user context and preferences, and outlines various challenges and strategies for maximizing user engagement. Key areas of focus include action selection, the importance of artwork diversity, and the potential for reinforcement learning to improve long-term user satisfaction.
![● Unbiased offline evaluation from explore data
Offline metric: Replay [Li et al, 2010]
Offline Take Fraction = 2 / 3
User 1 User 2 User 3 User 4 User 5 User 6
Random Assignment
Play?
Model Assignment](https://image.slidesharecdn.com/fernandoamatnetflix20180912recsys2018industrytalkimagepersonalization-181004223121/85/Artwork-Personalization-at-Netflix-Fernando-Amat-RecSys2018-17-320.jpg)
Artwork Personalization at Netflix Fernando Amat RecSys2018
- 2. Quickly help members discover content they’ll love
- 3. Global Members, Personalized Tastes 130 Million Members ~180 Countries
- 6. Artwork Optimization Goal: Recommend a personalized artwork or imagery for a title to help members decide if they will enjoy the title or not.
- 7. Intuition for Personalized Assets ● Emphasize themes through different artwork according to some context (user, viewing history, country, etc.) Preferences in genre
- 8. Intuition for Personalized Assets ● Emphasize themes through different artwork according to some context (user, viewing history, country, etc) Preferences in cast members
- 9. Bandit Algorithms Setting For each (user, show) request: ● Actions: set of candidate images available ● Reward: how many minutes did the user play from that impression ● Environment: Netflix homepage in user’s device ● Learner: its goal is to maximize the cumulative reward after N requests Learner Environment Action Reward Context
- 10. Numerous Variants ● Different Strategies: ε-Greedy, Thompson Sampling (TS), Upper Confidence Bound (UCB), etc. ● Different Environments: ○ Stochastic and stationary: Reward is generated i.i.d. from a distribution specific to the action. No payoff drift. ○ Adversarial: No assumptions on how rewards are generated. ● Different objectives: Cumulative regret, tracking the best expert ● Continuous or discrete set of actions, finite vs infinite ● Extensions: Varying set of arms, Contextual Bandits, etc.
- 11. Specific challenges ● Play attribution and reward assignment ○ Incremental effect of the image on top of recommender system ● Only one image per title can be presented ○ Although inherently it is a ranking problem Would you play because the movie is recommended or because of the artwork? Or both?
- 12. Specific challenges ● Change effect ○ Can changing images too often make users confused? Session 1 Session 2 Session 3 ... Session N Sequence A Sequence B
- 13. ● We have control over the set of actions ○ How many images per show ○ Image design ● What makes a good asset? ○ Representative (no clickbait) ○ Differential ○ Informative ○ Engaging Actions Personal (i.e. contextual)
- 14. Explore show? Choose Epsilon Greedy Example εprofile 1-εprofile εshow 1-εshow Personalized Image Image At Random
- 15. ● Learn a binary classifier per image to predict probability of play ● Pick the winner (arg max) Member (context) Features Image Pool Model 1 Winner arg max Model 2 Model 3 Model 4 Greedy Policy Example
- 16. Take Fraction Example: Luke Cage Take Fraction = 1 / 3 Play No play User A User B User C
- 17. ● Unbiased offline evaluation from explore data Offline metric: Replay [Li et al, 2010] Offline Take Fraction = 2 / 3 User 1 User 2 User 3 User 4 User 5 User 6 Random Assignment Play? Model Assignment
- 18. Offline Replay ● Context matters ● Artwork diversity matters ● Personalization wiggles around most popular images Lift in Replay in the various algorithms as compared to the Random baseline
- 19. Online results ● Rollout to our >130M member base ● Most beneficial for lesser known titles ● Compression from title -level offline metrics due to cannibalization between titles
- 21. Action selection orchestration ● Neighboring image selection influences result ● Title-level optimization is not enough Row A (diverse images) Row B (the microphone row) Stand-up comedy
- 22. Automatic image selection ● Generating new artwork is costly and time consuming ● Develop algorithm to predict asset quality from raw image
- 23. Long-term Reward: Road to RL ● Maximize long term reward: reinforcement learning ○ User long term joy rather than plays
- 24. Thank you. Fernando Amat ([email protected]) Blogpost We are hiring!