ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 1812.08944
99
23

Isotonic Regression in Multi-Dimensional Spaces and Graphs

21 December 2018
Hang Deng
Cun-Hui Zhang
ArXivPDFHTML
Abstract

In this paper we study minimax and adaptation rates in general isotonic regression. For uniform deterministic and random designs in [0,1]d[0,1]^d[0,1]d with d≥2d\ge 2d≥2 and N(0,1)N(0,1)N(0,1) noise, the minimax rate for the ℓ2\ell_2ℓ2​ risk is known to be bounded from below by n−1/dn^{-1/d}n−1/d when the unknown mean function fff is nondecreasing and its range is bounded by a constant, while the least squares estimator (LSE) is known to nearly achieve the minimax rate up to a factor (log⁡n)γ(\log n)^\gamma(logn)γ where nnn is sample size, γ=4\gamma = 4γ=4 in the lattice design and γ=max⁡{9/2,(d2+d+1)/2}\gamma = \max\{9/2, (d^2+d+1)/2 \}γ=max{9/2,(d2+d+1)/2} in the random design. Moreover, the LSE is known to achieve the adaptation rate (K/n)−2/d{1∨log⁡(n/K)}2γ(K/n)^{-2/d}\{1\vee \log(n/K)\}^{2\gamma}(K/n)−2/d{1∨log(n/K)}2γ when fff is piecewise constant on KKK hyperrectangles in a partition of [0,1]d[0,1]^d[0,1]d. Due to the minimax theorem, the LSE is identical on every design point to both the max-min and min-max estimators over all upper and lower sets containing the design point. This motivates our consideration of estimators which lie in-between the max-min and min-max estimators over possibly smaller classes of upper and lower sets, including a subclass of block estimators. Under a qqq-th moment condition on the noise, we develop ℓq\ell_qℓq​ risk bounds for such general estimators for isotonic regression on graphs. For uniform deterministic and random designs in [0,1]d[0,1]^d[0,1]d with d≥3d\ge 3d≥3, our ℓ2\ell_2ℓ2​ risk bound for the block estimator matches the minimax rate n−1/dn^{-1/d}n−1/d when the range of fff is bounded and achieves the near parametric adaptation rate (K/n){1∨log⁡(n/K)}d(K/n)\{1\vee\log(n/K)\}^{d}(K/n){1∨log(n/K)}d when fff is KKK-piecewise constant. Furthermore, the block estimator possesses the following oracle property in variable selection: When fff depends on only a subset SSS of variables, the ℓ2\ell_2ℓ2​ risk of the block estimator automatically achieves up to a poly-logarithmic factor the minimax rate based on the oracular knowledge of SSS.

View on arXiv
Comments on this paper