arXiv：cs.LG· Parham Nooranbakht, Seyed Mahdi Basiri Azad, Joschka B\"odecker·· 6 小时前
Selective Growing Q-Networks（SGQN）：解耦 Q-Learning 中自适应动作网格的分析
Selective Growing Q-Networks (SGQN): An Analysis of Adaptive Action Grids in Decoupled Q-Learning
AI 导读
研究者提出 Selective Growing Q-Networks（SGQN），在解耦 Q-Learning 中用自适应动作网格替代均匀增长：在高 Q 值分箱旁增加分箱、剪除低 Q 值分箱，并以 epsilon-greedy 选动作。
正文
Abstract:Value-based Q-learning for continuous control can discretise the action space, but the resolution of the discretisation trades early exploration against final precision. Growing Q-Networks (GQN) grow the resolution uniformly from coarse to fine. We introduce Selective Growing Q-Networks (SGQN), which keeps decoupled Q-learning and replaces uniform growth with an adaptive action grid: bins are added next to bins with high estimated Q-value, bins with low Q-value are pruned, and actions are chosen epsilon-greedily. On eight DeepMind Control Suite (DMControl) tasks with 2 to 12 action dimensions, SGQN has the highest mean final return among the compared methods on four, is within 1.1% of the best method on three more, and is 3.4% below Bang-Bang MPO on Walker Run, using on average 7.5 of 9 available bins per dimension. We then test whether the adaptive grid explains these results, and find that it does not. Growth activates every candidate bin and behaves like uniform growth, and removing pruning changes the return by at most 1.1%. Two other choices matter: starting from a coarse grid, since starting with all 9 bins produces collapsed seeds, and epsilon-greedy selection, since an annealed softmax lowers the return by 23% to 100%. SGQN is insensitive to its growth and pruning hyperparameters. It also differs from the GQN and DecQN baselines in other respects, notably one gradient update per environment step instead of one per eight, which we did not control for, so we do not attribute its advantage to a single component. We present SGQN as a competitive value-based method and as an analysis of which ingredients of a growing action grid matter.
| Comments: | 19 pages, 6 figures, 6 tables. Under review at TMLR |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.20463 [cs.LG] |
| (or arXiv:2610.20463v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.20463 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Parham Nooranbakht [view email]
[v1]
Wed, 14 Oct 2026 16:45:10 UTC (1,702 KB)
来源：arXiv：cs.LG · arxiv.org