Completed
Particle Thompson Sampling (PTS) [KBKTC15] • Probabilistic Matrix Factorization framework • Particle filtering for online Bayesian parameter estimation • Thompson Sampling for exploration
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
KDD 2020- Learning by Exploration-Part 2
Automatically move to the next video in the Classroom when playback concludes
- 1 Intro
- 2 Build independent LinUCB for each user? . Cold start challenge • Users are not independent
- 3 Connected users are assumed to share similar model parameters • Graph Laplacan based regularization upon ridge regression to model dependency
- 4 Graph Laplacian based regularization upon ridge regression to model dependency • Encode graph Laplaclan in context formulate as a di dimensional LIUCB
- 5 Social influence among users. content and opinion sharing in social network W • Reward: weighted average of expected reward among friends
- 6 Adaptively cluster users into groups by keep removing edges
- 7 item clustering • Each item cluster is associated with its own user clustering
- 8 Context-dependent clustering . For current user i, find neighboring user set /for every candidate item X. . Then aggregate the history rewards/ predictions within the user cluster.
- 9 Particle Thompson Sampling (PTS) [KBKTC15] • Probabilistic Matrix Factorization framework • Particle filtering for online Bayesian parameter estimation • Thompson Sampling for exploration
- 10 Alternating Least Squares for optimization • Exploration considers uncertainty from two factors
- 11 Leverage historical data to warm start model, reduce the need of exploration
- 12 What is the problem-related (structure-related) regret lower bound . Eg, user dependency structure, low rank, offline data • Did current algorithms fully utilize the information in problem structure?
