Workspace sync: auto commit 2026-04-23 12:02:11
This commit is contained in:
23
wiki/concepts/Genetic-Algorithm.md
Normal file
23
wiki/concepts/Genetic-Algorithm.md
Normal file
@@ -0,0 +1,23 @@
|
||||
---
|
||||
title: "Genetic Algorithm"
|
||||
type: concept
|
||||
tags: []
|
||||
sources: []
|
||||
last_updated: 2026-04-25
|
||||
---
|
||||
|
||||
# Genetic Algorithm
|
||||
|
||||
## 定义
|
||||
遗传算法——传统机器学习中基于自然选择和遗传机制的优化算法,是[[Tree-of-Thoughts]]和[[Knock-out-Pattern]]的ML理论根源。
|
||||
|
||||
## 核心要素
|
||||
1. **遗传表示**(Genetic Representation):解决方案域的编码(模型+上下文)
|
||||
2. **适应度函数**(Fitness Function):评估解决方案质量的函数(淘汰赛裁判)
|
||||
|
||||
## 在多智能体系统中的应用
|
||||
- [[Knock-out-Pattern]]是遗传算法的精简实现——将适应度函数替换为验证器(Validator)
|
||||
- [[Tree-of-Thoughts]]通过验证器持续筛选Agent分支,可结合赢家的特征重组生成新Agent
|
||||
|
||||
## 来源
|
||||
- [[multi-agent-system-reliability]]
|
||||
Reference in New Issue
Block a user