MANAGING METHOD FOR MULTI-RESOURCE BASED REINFORCEMENT LEARNING USING ATTENTION TECHNIQUE

摘要

A method for multi-resource management based on reinforcement learning to which an attention technique is applied according to an embodiment of the present invention is a first step of receiving a component related to a task or machine in a cluster by performing a local encoding conversion, calculated through the local encoding conversion A second step of performing global encoding conversion by applying an attention technique to the local encoding vector, a third step of calculating a score function regarding the importance of a task using the conversion result of the global encoding conversion, and the second step and and after repeating the third step, mapping the optimized task and machine to each other.