Deterministic and stochastic convergence properties of AIMD algorithms with nonlinear back-off functions.
Martin J. CorlessRobert ShortenPublished in: Autom. (2012)
Keyphrases
- stochastic approximation
- learning algorithm
- computational cost
- orders of magnitude
- significant improvement
- theoretical analysis
- black box
- monte carlo sampling
- randomized algorithms
- theoretical justification
- deterministic finite automata
- rapid convergence
- weight update
- nonlinear functions
- optimization methods
- benchmark datasets
- computationally efficient
- multi objective