基于算法运行系统的进化学习算法运行方法、系统、介质
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- GUANGDONG INST OF INTELLIGENT SCI & TECH
- Filing Date
- 2024-08-27
- Publication Date
- 2026-07-17
AI Technical Summary
Existing evolutionary learning algorithm systems face the problem of insufficient computational accuracy in low-precision AI accelerator cards and their cluster systems, making them difficult to apply effectively.
By dynamically allocating resources between high-precision computing devices and AI accelerators, the target computing device is determined. Then, the evolutionary learning algorithm is optimized in a low-precision environment using gradient-free optimization strategies and evolutionary operators to generate an initial solution and evaluate its fitness, ultimately obtaining the target solution.
While ensuring computational accuracy, it expands the hardware application scope of evolutionary learning algorithms, enabling them to run stably on low-precision AI accelerator cards, thereby improving computational efficiency and throughput.
Smart Images
Figure CN119150956B_ABST