The Noisy Power Method: A Meta Algorithm with Applications.
Moritz HardtEric PricePublished in: NIPS (2014)
Keyphrases
- significant improvement
- high accuracy
- cost function
- experimental evaluation
- dynamic programming
- optimization algorithm
- computational cost
- detection algorithm
- improved algorithm
- clustering method
- objective function
- segmentation algorithm
- noisy data
- synthetic and real images
- preprocessing
- theoretical analysis
- optimization method
- computational complexity
- segmentation method
- classification method
- computational efficiency
- learning algorithm
- convergence rate
- input data
- recognition algorithm
- high efficiency
- detection method
- tree structure
- matching algorithm
- em algorithm
- prior information
- experimental study
- estimation algorithm
- single pass
- combinatorial optimization
- mapping function
- classification algorithm
- similarity measure
- energy function
- support vector machine svm
- reconstruction method
- selection algorithm
- probabilistic model
- mathematical model
- pairwise
- incomplete data
- k means
- image matching
- particle swarm optimization
- region of interest
- highly efficient
- test images
- lower bound
- kalman filter
- decomposition method
- support vector machine
- np hard
- expectation maximization
- search space
- bayesian networks
- noise free