Reducing Variance in Gradient Bandit Algorithm using Antithetic Variates Method.
Sihao YuJun XuYanyan LanJiafeng GuoXueqi ChengPublished in: SIGIR (2018)
Keyphrases
- preprocessing
- dynamic programming
- high accuracy
- detection algorithm
- objective function
- improved algorithm
- experimental evaluation
- computational cost
- clustering method
- recognition algorithm
- detection method
- optimization algorithm
- computationally efficient
- significant improvement
- computational complexity
- cost function
- synthetic and real images
- theoretical analysis
- support vector machine svm
- segmentation algorithm
- k means
- matching algorithm
- classification algorithm
- input data
- prediction error
- similarity measure
- tree structure
- classification method
- gradient vectors
- segmentation method
- estimation algorithm
- gradient method
- gradient information
- learning algorithm
- convergence rate
- optimization method
- selection algorithm
- probabilistic model
- gradient orientation
- energy function
- expectation maximization
- reconstruction method
- genetic algorithm
- steepest ascent
- principal components
- correlation coefficient
- covariance matrix
- simulated annealing
- edge detection
- optimal solution
- minimum variance