Deep Predictor-Corrector Networks for Robust Parameter Estimation in Non-autonomous System with Discontinuous Inputs
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866908884213432320 |
|---|---|
| author | Gu, Gyeongwan Hyun, Jinwoo Jo, Hyeontae Kim, Jae Kyoung |
| author_facet | Gu, Gyeongwan Hyun, Jinwoo Jo, Hyeontae Kim, Jae Kyoung |
| contents | Learning under non-smooth objectives remains a fundamental challenge in machine learning, as abrupt changes in conditioning variables can induce highly non-smooth loss landscapes and destabilize optimization. This difficulty is particularly pronounced in non-autonomous dynamical systems driven by discontinuous inputs, where widely used optimization methods, including recent neural smoothing approaches, exhibit unreliable convergence or strong hyperparameter sensitivity. To address this issue, we propose Deep Predictor-Corrector Networks (DePCoN), a multi-scale learning framework that stabilizes optimization by learning scale-consistent parameter update rules across a hierarchy of smoothed inputs. Rather than treating smoothing as a fixed preprocessing choice, DePCoN integrates smoothing into the learning dynamics itself through a learned predictor-corrector mechanism. Across biological and ecological benchmarks with discontinuous inputs, DePCoN consistently achieves more robust and faster convergence than existing methods while substantially reducing sensitivity to hyperparameter choices. Beyond dynamical systems, our approach provides a general learning principle for stabilizing optimization under non-smooth objectives. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_12965 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Deep Predictor-Corrector Networks for Robust Parameter Estimation in Non-autonomous System with Discontinuous Inputs Gu, Gyeongwan Hyun, Jinwoo Jo, Hyeontae Kim, Jae Kyoung Optimization and Control 34C60, 92B05, 68T07, 93C15, 65K10 Learning under non-smooth objectives remains a fundamental challenge in machine learning, as abrupt changes in conditioning variables can induce highly non-smooth loss landscapes and destabilize optimization. This difficulty is particularly pronounced in non-autonomous dynamical systems driven by discontinuous inputs, where widely used optimization methods, including recent neural smoothing approaches, exhibit unreliable convergence or strong hyperparameter sensitivity. To address this issue, we propose Deep Predictor-Corrector Networks (DePCoN), a multi-scale learning framework that stabilizes optimization by learning scale-consistent parameter update rules across a hierarchy of smoothed inputs. Rather than treating smoothing as a fixed preprocessing choice, DePCoN integrates smoothing into the learning dynamics itself through a learned predictor-corrector mechanism. Across biological and ecological benchmarks with discontinuous inputs, DePCoN consistently achieves more robust and faster convergence than existing methods while substantially reducing sensitivity to hyperparameter choices. Beyond dynamical systems, our approach provides a general learning principle for stabilizing optimization under non-smooth objectives. |
| title | Deep Predictor-Corrector Networks for Robust Parameter Estimation in Non-autonomous System with Discontinuous Inputs |
| topic | Optimization and Control 34C60, 92B05, 68T07, 93C15, 65K10 |
| url | https://arxiv.org/abs/2603.12965 |