Stochasticity helps to navigate rough landscapes: comparing gradient-descent-based algorithms in the phase retrieval problem.
Francesca MignaccoPierfrancesco UrbaniLenka ZdeborováPublished in: CoRR (2021)
Keyphrases
- learning algorithm
- computational cost
- orders of magnitude
- machine learning algorithms
- theoretical analysis
- information retrieval
- benchmark datasets
- computationally efficient
- conjugate gradient
- convergence rate
- times faster
- computational efficiency
- optimization problems
- simulated annealing
- image database
- relevance feedback
- data structure
- image segmentation
- metadata