First page Back Continue Last page Overview Graphics
Space Alternating Generalized Expectation (SAGE) Maximization algorithm provides an iterative approach to parameter estimation when direct maximization of the likelihood function may be infeasible.
Complexity is less in those applications where convergence rate is fast.
SAGE is a low complexity algorithm for DSL as only one iteration is required for convergence.
Starting from some initial guess, each iteration consists of two steps
- E-step (Expectation step)
- M-step (Maximization step)