Research Article
An Improved Adam Optimization Algorithm Combining Adaptive Coefficients and Composite Gradients Based on Randomized Block Coordinate Descent
Algorithm 2
: Randomized block coordinate descent (RBC).
| Input: ( is a N-dimensional set of real numbers) | | Output: | (1) | Initialize parameters | (2) | For t = 1 to T do | (3) | t = t + 1 | (4) | Select coordinates evenly and randomly | (5) | Update where are Lipschitz constants | (6) | End For |
|