Beyond Online Balanced Descent: An Optimal Algorithm for Smoothed Online Optimization.
Gautam GoelYiheng LinHaoyuan SunAdam WiermanPublished in: NeurIPS (2019)
Keyphrases
- dynamic programming
- optimization algorithm
- optimal solution
- learning algorithm
- passive aggressive
- online learning
- probabilistic model
- computational complexity
- optimal design
- globally optimal
- real time
- worst case
- computational cost
- cost function
- closed form
- exhaustive search
- np hard
- preprocessing
- high accuracy
- stochastic gradient
- optimization model
- k means
- locally optimal
- constrained optimization
- matching algorithm
- optimal parameters
- online algorithms
- max min
- optimization method
- detection algorithm
- segmentation algorithm
- combinatorial optimization
- significant improvement
- search space
- optimal path
- genetic algorithm
- approximately optimal
- optimality criterion
- regret bounds
- feature selection
- optimization strategies
- objective function
- multi objective
- experimental evaluation
- times faster
- optimization process
- improved algorithm
- convergence rate