First-order algorithm with 𝒪(ln(1/ε)) convergence for ε-equilibrium in two-person zero-sum games.
Andrew GilpinJavier PeñaTuomas SandholmPublished in: Math. Program. (2012)
Keyphrases
- iterative algorithms
- detection algorithm
- dynamic programming
- learning algorithm
- convergence rate
- experimental evaluation
- computational cost
- recognition algorithm
- optimization algorithm
- expectation maximization
- high accuracy
- np hard
- faster convergence
- theoretical analysis
- higher order
- cost function
- significant improvement
- preprocessing
- similarity measure
- k means
- search space
- monte carlo
- data structure
- optimal solution
- objective function