20260617:目前有914 页
This commit is contained in:
57
concepts/signature.md
Normal file
57
concepts/signature.md
Normal file
@@ -0,0 +1,57 @@
|
||||
---
|
||||
title: "签名 (Signature of Paths)"
|
||||
created: 2026-06-17
|
||||
updated: 2026-06-17
|
||||
type: concept
|
||||
tags: [rough-paths, stochastic-analysis, representation-learning, signature-methods]
|
||||
sources: [raw/papers/schmocker-weighted-uat-2026.md]
|
||||
confidence: high
|
||||
---
|
||||
|
||||
# 签名 (Signature of Paths)
|
||||
|
||||
签名是**路径的泛函多项式基**——可以理解为路径空间上的"Taylor 展开"或"非交换多项式"。[[weighted-uat-manifolds|Schmocker & Teichmann (2026)]] 证明签名的线性函数可以逼近任意路径泛函(含导数)。
|
||||
|
||||
## 定义
|
||||
|
||||
对路径 `X : [0,T] → R^d`,其截断到 K 阶的签名为:
|
||||
|
||||
```
|
||||
S^K(X) = (1, S^1, S^2, ..., S^K)
|
||||
|
||||
其中 S^k(X) = ∫_{0≤t_1<...<t_k≤T} dX_{t_1} ⊗ ... ⊗ dX_{t_k}
|
||||
```
|
||||
|
||||
一阶:增量向量 `X_T - X_0`
|
||||
二阶:Lévy 面积 / 路径围成的面积
|
||||
高阶:高阶迭代积分
|
||||
|
||||
## 核心性质
|
||||
|
||||
1. **泛函逼近**:任意连续路径泛函可被签名的线性函数任意逼近(类似多项式逼近连续函数)
|
||||
2. **对重参数化不变**:签名对时间重参数化不变(捕获几何信息)
|
||||
3. **特征唯一性**:rough paths 被其签名唯一确定
|
||||
|
||||
## 论文中的推广
|
||||
|
||||
Schmocker & Teichmann 证明了加权版本的 Signature UAT:
|
||||
|
||||
- 线性签名函数在**加权路径空间**中稠密
|
||||
- **同时逼近泛函值及其方向导数**
|
||||
- 逼近在**整个路径空间**上成立,不含紧限制
|
||||
|
||||
## Rough Path 理论
|
||||
|
||||
签名是 [[rough-path-theory|粗糙路径理论]](Lyons, 1998)的中心概念——使 Itô 积分和 SDE 理论可在确定性路径设定中严格表述。
|
||||
|
||||
## 机器学习的应用
|
||||
|
||||
- **Signature Methods**:将不规则时序数据映射为固定长度特征
|
||||
- **Log-Signature**:签名的对数,更低维度
|
||||
- **Neural SDEs / Deep Signature Transforms**
|
||||
|
||||
## 参考
|
||||
|
||||
- [[rough-path-theory|粗糙路径理论]]
|
||||
- [[non-anticipative-functionals|非预期泛函]]
|
||||
- [[weighted-uat-manifolds|论文原文]]
|
||||
Reference in New Issue
Block a user