Learning Beyond Optimization: Stress-Gated Dynamical Regime Regulation in Autonomous Systems
Sheng Ran · 2026
Read Learning Beyond Optimization: Stress-Gated Dynamical Regime Regulation in Autonomous Systems by Sheng Ran (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-18581.txt.
Description
Sheng Ran proposes learning without a set goal. His 2026 paper notes modern machine learning reduces to tuning parameters against a scalar goal, highly successful where goals are fixed and criteria set. But for true autonomy over long horizons and changing contexts, objectives may turn ill-defined, shifting, or missing.
The paper asks how a system can judge whether its internal dynamics are healthy or pathological without a set goal, and how it should regulate structural change without outside supervision. Ran proposes a dynamic framework where the system checks the health of its own dynamics and regulates plasticity. A two-timescale architecture separates fast evolution from slow adaptation, linked through an internally generated stress variable tracking signs of dysfunction. Structural change triggers as needed, not always, and a minimal toy model demonstrates it.
Catalog record
| CREATOR | Sheng Ran |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-18581