| _version_ | 1866902106792787968 |
|---|---|
| author | Souilah, Rachik |
| author_facet | Souilah, Rachik |
| contents | <p>We address the vulnerability of Deep Neural Networks (DNNs) to catastrophic forgetting under<br>severe non-stationary data drifts (trauma). Conventional safety mechanisms, such as Gradient Clip-<br>ping, constrain the instantaneous magnitude of updates but fail to prevent the cumulative erosion of<br>prior representations under sustained stress. This paper introduces Neural Guardian, a meta-control<br>layer that enforces a long-term energetic invariant on the learning process. Unlike optimization strate-<br>gies that seek to minimize loss, Neural Guardian operates as an independent safety kernel: it moni-<br>tors an energetic proxy derived from the cumulative gradient norm and enforces a Bounded Transient<br>Release (BTR) logic. When cumulative stress exceeds a critical threshold, the kernel asserts a mul-<br>tiplicative veto (α → 0) on the optimizer, imposing a mandatory dissipation period. Experimental<br>results on a TimesFM proxy demonstrate that while Gradient Clipping only delays performance<br>collapse, Neural Guardian successfully preserves nominal task performance (Performance Decay<br>Rate ≈ 0) by strictly refusing to integrate incompatible distributional shifts. This establishes Neural<br>Guardian not as an optimizer, but as a necessary boundary condition for lifetime stability in adaptive<br>systems.</p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_18448348 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Neural Guardian: Energetic Admissibility as a Defense Against Catastrophic Forgetting in Deep Learning Souilah, Rachik Catastrophic Forgetting Deep Learning Safety Energetic Admissibility Neural Guardian Bounded Transient Release Meta-Control Non-Stationary Dynamics TimesFM <p>We address the vulnerability of Deep Neural Networks (DNNs) to catastrophic forgetting under<br>severe non-stationary data drifts (trauma). Conventional safety mechanisms, such as Gradient Clip-<br>ping, constrain the instantaneous magnitude of updates but fail to prevent the cumulative erosion of<br>prior representations under sustained stress. This paper introduces Neural Guardian, a meta-control<br>layer that enforces a long-term energetic invariant on the learning process. Unlike optimization strate-<br>gies that seek to minimize loss, Neural Guardian operates as an independent safety kernel: it moni-<br>tors an energetic proxy derived from the cumulative gradient norm and enforces a Bounded Transient<br>Release (BTR) logic. When cumulative stress exceeds a critical threshold, the kernel asserts a mul-<br>tiplicative veto (α → 0) on the optimizer, imposing a mandatory dissipation period. Experimental<br>results on a TimesFM proxy demonstrate that while Gradient Clipping only delays performance<br>collapse, Neural Guardian successfully preserves nominal task performance (Performance Decay<br>Rate ≈ 0) by strictly refusing to integrate incompatible distributional shifts. This establishes Neural<br>Guardian not as an optimizer, but as a necessary boundary condition for lifetime stability in adaptive<br>systems.</p> |
| title | Neural Guardian: Energetic Admissibility as a Defense Against Catastrophic Forgetting in Deep Learning |
| topic | Catastrophic Forgetting Deep Learning Safety Energetic Admissibility Neural Guardian Bounded Transient Release Meta-Control Non-Stationary Dynamics TimesFM |
| url | https://doi.org/10.5281/zenodo.18448348 |