A Provably Efficient Algorithm for Training Deep Networks
Roi LivniShai Shalev-ShwartzOhad ShamirPublished in: CoRR (2013)
Keyphrases
- computational complexity
- preprocessing
- computationally efficient
- dynamic programming
- k means
- cost function
- optimization algorithm
- worst case
- recognition algorithm
- optimal solution
- highly efficient
- experimental evaluation
- high efficiency
- detection algorithm
- improved algorithm
- expectation maximization
- times faster
- matching algorithm
- single pass
- theoretical guarantees
- simulated annealing
- np hard
- search space
- genetic algorithm
- neural network
- pruning strategy
- feed forward neural networks
- learning algorithm
- social networks
- training algorithm
- image segmentation
- convergence rate
- high accuracy
- clustering method
- linear programming
- input data
- segmentation algorithm
- theoretical analysis