Files
myWiki/concepts/isotonic-regression.md

36 lines
1.3 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

---
title: "Isotonic Regression"
created: 2026-06-18
updated: 2026-06-18
type: concept
tags: ["optimization", "statistics", "convex-relaxation"]
sources: ["https://arxiv.org/abs/2602.08585"]
---
# Isotonic Regression
## 定义
Isotonic Regression保序回归是一种约束优化方法在保持数据点单调性约束非降或非增的前提下拟合一个序列。在 LU-KV 的 [[convex-hull-relaxation]] 中PVAAPool Adjacent Violators Algorithm被用于对离散损失序列做保序回归将其投影为单调非增序列从而得到凸化代用损失。
## PAVA 算法
PAVAPool Adjacent Violators Algorithm是最经典的保序回归求解器
1. 遍历序列检测违反单调约束的相邻元素violators
2. 将 violators 合并为 block赋值为 block 内元素的平均值
3. 重复直到序列满足单调约束
在 LU-KV 中PAVA 将非单调的边际递减量 d(i) 投影为单调非增的 d̆(i)。
## 相关概念
- [[convex-hull-relaxation]] — 在 LU-KV 中通过保序回归实现凸松弛
- [[global-combinatorial-optimization]] — 保序回归使贪心解达到最优
- [[tang-lukv|LU-KV]] — 使用 PAVA 的论文
## 参考
- Barlow & Brunk (1972) — PAVA 算法的经典文献
- [[tang-lukv|LU-KV]] (Tang et al., ICML 2026) — 附录 A.2