Metraipolis › archive › dc_ffc49ad282c4

DAMPER: Return-Prioritized Gradient Control for Smooth Policies — arXiv:2609.38903

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.38903
License (as archived)
CC0-1.0
Archived
2026-10-07T12:44:32.969Z
SHA-256 of the archived text
06f05bed839a8d441acc6e8e527fd7cfe246dd27ef8314ca288e2832d0eace2e
Stable URL of this page
https://metraipolis.com/documento/dc_ffc49ad282c4

How to cite

DAMPER: Return-Prioritized Gradient Control for Smooth Policies — arXiv:2609.38903 — arXiv (en), https://arxiv.org/abs/2609.38903 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ffc49ad282c4 (Metraipolis archive dc_ffc49ad282c4, sha256 06f05bed839a8d441acc6e8e527fd7cfe246dd27ef8314ca288e2832d0eace2e)
BibTeX
@misc{dcffc49ad282c4,
  title = {DAMPER: Return-Prioritized Gradient Control for Smooth Policies — arXiv:2609.38903},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.38903},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ffc49ad282c4; sha256 06f05bed839a8d441acc6e8e527fd7cfe246dd27ef8314ca288e2832d0eace2e},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38903. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Actor-critic methods achieve strong performance in continuous control, but their policies can produce highly oscillatory actions. A common remedy is to add auxiliary smoothness losses. However, their contribution can be negligible when their gradients are small relative to the native actor gradient. Moreover, existing methods often combine multiple auxiliary losses, complicating loss balancing without necessarily improving the return-smoothness trade-off. We introduce DAMPER (Direction-Aware Magnitude-Controlled Projection with Explicit Return Priority), which combines the native actor gradient with a temporal-consistency gradient through conflict-conditioned projection and adaptive magnitude control. It removes the auxiliary component opposing the actor gradient and scales the retained temporal direction relative to the actor gradient norm, preserving positive alignment with the native actor gradient. Experiments with TD3 and SAC on six continuous-control tasks show reduced action oscillation relative to the native agents in all 12 task-backbone pairs and the best oscillation score among the compared methods in eight, with task-dependent return trade-offs.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.38903 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T12:14:13.973Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).