Deep reinforcement learning cooperative scheduling method and device for heterogeneous computing resources

• Main Authors: XU CHENXI, SHENTU LEIXUAN, LIU WEIWEI, WANG YUE, ZHENG MIN, HUANG YUANPING
• Format: Patent
• Language: Chinese; English
• Published: 19.04.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention relates to the technical field of artificial intelligence computing power resource scheduling, and discloses a deep reinforcement learning collaborative scheduling method and device for heterogeneous computing resources, and the method comprises the steps: obtaining a deep learning task and heterogeneous computing environment information, the heterogeneous computing environment being a cloud edge collaborative environment; constructing a task scheduling model according to the deep learning task and the heterogeneous computing environment information; and adopting a task scheduling algorithm based on deep reinforcement learning to schedule the deep learning task according to the task scheduling model. According to the method, the aim of considering the deep learning task and heterogeneous computing environment information in the scheduling process is fulfilled, and the deep learning task is scheduled by using the constructed task model and a task scheduling algorithm based on deep reinforcement l