Large Stepsize Gradient Descent for Non-Homogeneous Two-Layer Networks: Margin Improvement and Fast Optimization.
Yuhang CaiJingfeng WuSong MeiMichael LindseyPeter L. BartlettPublished in: CoRR (2024)
Keyphrases
- step size
- stochastic gradient descent
- cost function
- quasi newton
- faster convergence
- convergence rate
- objective function
- optimization algorithm
- steepest descent method
- optimization problems
- convergence speed
- global optimization
- optimization methods
- loss function
- temporal difference
- global optimum
- image processing
- approximate dynamic programming
- social networks
- multi objective
- feature extraction
- search direction
- genetic algorithm