---
格式版本: 2
标题: "From Privileged Control to Deployable Adaptation:Fusing Mechanism-Guided Task Reduction with Learned Behavior"
原文链接: "https://arxiv.org/abs/2608.10453"
发布日期: "2026-08-11"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Tue, 11 Aug 2026 04:06:47 UTC (3,466 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-12T11:17:39+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-12T11:13:54+08:00"
入库时间: "2026-08-12T03:17:39.296Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=deployment&searchtype=all"
匹配关键词:
  - "deployment"
  - "performance"
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 10
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "该资料为控制系统学术论文，内容不涉及超节点/AI Rack/机柜级AI基础设施、GPU、液冷、供电、互连等任何项目主题，属于明显无关。"
AI质检模型: "ali-deepseek-v4-flash"
AI质检时间: "2026-08-12T11:18:03+08:00"
AI主题相关性: 0
AI来源权威性: 5
AI新颖性: 5
AI技术细节: 0
AI商业部署信号: 0
AI完整性: 0
采集批次: "2026年8月12日2点04分20秒"
采集批次ID: "20260812-020420-289"
去重键: "https://arxiv.org/abs/2608.10453"
---

## Electrical Engineering and Systems Science > Systems and Control

## Title:From Privileged Control to Deployable Adaptation:Fusing Mechanism-Guided Task Reduction with Learned Behavior

Authors:[Xitong Niu](https://arxiv.org/search/eess?searchtype=author&query=Niu,+X), [Peifeng Hui](https://arxiv.org/search/eess?searchtype=author&query=Hui,+P), [Zheyong Jiang](https://arxiv.org/search/eess?searchtype=author&query=Jiang,+Z), [Yuan Gao](https://arxiv.org/search/eess?searchtype=author&query=Gao,+Y), [Chuanlin Zhang](https://arxiv.org/search/eess?searchtype=author&query=Zhang,+C)

[View PDF](https://arxiv.org/pdf/2608.10453) [HTML (experimental)](https://arxiv.org/html/2608.10453v1)

> Abstract:Simultaneous input-gain variation and large additive disturbance create a control problem in which a fixed observer or nominal controller may be unable to reproduce the performance of a regime-aware design. We study a training--deployment asymmetry: during simulation or commissioning, an expert controller is allowed to use the known gain and disturbance, whereas the deployed controller can use only the reference and measured states. Directly imitating expert actions is generally unsafe because the same instantaneous student observation may correspond to different privileged regimes and hence different expert actions. We propose a mechanism-guided transfer route rather than a new neural architecture. An exact sampled-data identity removes the additive disturbance from the expert law and reduces learning to a task-relevant inverse input gain inferred from causal state history. The latent target is reconstructed from expert actions and deployment-visible trajectories, so the true plant parameter is not required as a student label. A common-quadratic certificate is derived for the actual augmented sampled recursion, followed by explicit residual, coverage, switching, noise, and saturation qualifications. A parameter-regime scan shows that the nominal observer's error grows sharply as $a$ decreases and that the next gain above the best non-failing tuning diverges for every tested $a<1$. Direct action networks also fail in closed loop despite moderate offline error, whereas the structured student remains close to the privileged expert and reduces tracking RMSE by about 69\\% relative to the tuned observer in unseen 60-s trials. The contribution is an interpretable design perspective for turning privileged multi-regime control knowledge into a deployable adaptive controller, together with conditions under which the transfer is meaningful.

| Comments: |  |
| --- | --- |
| Subjects: | Systems and Control (eess.SY) |
| Cite as: | [arXiv:2608.10453](https://arxiv.org/abs/2608.10453) \[eess.SY\] |
|  | (or [arXiv:2608.10453v1](https://arxiv.org/abs/2608.10453v1) \[eess.SY\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.10453](https://doi.org/10.48550/arXiv.2608.10453) |

## Submission history

From: Xitong Niu \[[view email](https://arxiv.org/show-email/68b7e65c/2608.10453)\]  
**\[v1\]** Tue, 11 Aug 2026 04:06:47 UTC (3,466 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.10453) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))