The Role of Momentum Parameters in the Optimal Convergence of Adaptive Polyak's Heavy-ball Methods.
Wei TaoSheng LongGaowei WuQing TaoPublished in: CoRR (2021)
Keyphrases
- significant improvement
- optimal solution
- dynamic programming
- parameter values
- maximum likelihood
- worst case
- adaptive systems
- convergence speed
- optimization methods
- parameter settings
- machine learning methods
- benchmark datasets
- vision system
- neural network
- least squares
- computational cost
- k means
- image segmentation
- clustering algorithm
- data mining