A Fast Bandit Algorithm for Recommendation to Users With Heterogenous Tastes.
Pushmeet KohliMahyar SalekGreg StoddardPublished in: AAAI (2013)
Keyphrases
- learning algorithm
- detection algorithm
- dynamic programming
- objective function
- preprocessing
- recommender systems
- k means
- computational complexity
- significant improvement
- cost function
- segmentation algorithm
- high accuracy
- computational cost
- np hard
- decision trees
- optimization algorithm
- user preferences
- experimental evaluation
- neural network
- information overload
- recognition algorithm
- times faster
- contextual bandit
- similarity measure
- expectation maximization
- collaborative filtering
- worst case
- search space
- reinforcement learning