Task scheduling method and related product
By accumulating the rewards of reinforcement learning models and preset algorithms in task scheduling and updating the model, the problem of reinforcement learning models training oscillations and difficulty converging in the fusion algorithm environment is solved, and a more stable task scheduling process is achieved.
CN119690622APending Publication Date: 2025-03-25CAMBRIAN (KUNSHAN) INFORMATION TECH CO LTD
0 Cites 0 Cited by
Patent Information
- Application Number
- CN202411814455.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-10
- Publication Date
- 2025-03-25
AI Technical Summary
Technical Problem
When integrating reinforcement learning scheduling algorithms and other preset algorithms for task scheduling, reinforcement learning model training oscillates and is difficult to converge.
Method used
By obtaining the rewards of reinforcement learning models and preset algorithms in each scheduling round, cumulative rewards and update the reinforcement learning model to stabilize its training process.
Benefits of technology
It effectively solves the problem of oscillation and difficulty in convergence training of reinforcement learning models, so that the model can more stably perceive the environmental fluctuations brought by different algorithms and improves the efficiency of task scheduling.
✦ Generated by Eureka AI based on patent content.
Abstract
The invention discloses a task scheduling method and a related product. The task scheduling method disclosed by the disclosure comprises the steps of obtaining at least one first reward obtained by reinforcement learning scheduling rounds of task scheduling based on a reinforcement learning model in current iteration; obtaining at least one second reward obtained by a preset algorithm scheduling round of task scheduling based on a preset algorithm in the current iteration; according to the at least one first reward and the at least one second reward, obtaining an accumulated reward of a plurality of reinforcement learning scheduling rounds in the current iteration; and updating the reinforcement learning model according to the accumulated reward so as to perform task scheduling by using the updated reinforcement learning model in the next iteration. According to the method, the reinforcement learning model is updated according to the accumulated reward obtained by the reinforcement learning scheduling algorithm and the preset algorithm, so that the reinforcement learning model can sense environmental fluctuation caused by different algorithms, and the convergence of the reinforcement learning model is improved.
Need to check novelty before this filing date? Find Prior Art