发表机构
University of Neuchâtel; Information Management Institute; Computer Science Department; University College Dublin（纳沙泰尔大学; 信息管理研究所; 计算机科学系; 都柏林大学学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对 ES-HyperNEAT 超参数配置易停滞浪费计算资源的问题，提出基于早期适应度轨迹二分类的早停规则（G*=3，T*=0.140），在验证集上 F1=0.872，节省 41.6% 计算成本，方法可推广至其他进化算法。
AI 中文摘要
大多数可进化基板 HyperNEAT（ES-HyperNEAT）的超参数配置产生的网络停滞在随机猜测性能水平，浪费计算资源。我们将早停问题表述为对早期适应度轨迹的二分类问题：对于每次试验，我们计算每代最佳适应度的累积中位数，并将其与通过最大化初始 90 次试验数据集上的 F1 分数得出的阈值进行比较。由此得到的规则（代数 G* = 3，阈值 T* = 0.140）在 180 次独立验证试验中达到 F1 = 0.872，保留了超过 90% 的成功试验，同时将计算成本削减了 41.6%。与 Hyperband 相比，我们的领域特定规则效率高出 64%，且平均适应度更高，尽管 Hyperband 偶尔能发现更高的峰值解。在收敛的搜索种群上，该规则变得过于激进（召回率 31.1%），这促使我们采用自适应阈值。具体阈值是 ES-HyperNEAT 特有的，但该方法论——从适应度动态分类中推导停止标准——适用于其他具有易停滞超参数空间的进化算法。
英文摘要
Most hyperparameter configurations for Evolvable-Substrate HyperNEAT (ES-HyperNEAT) produce networks that stagnate at random-guessing performance, wasting computational resources. We frame early stopping as binary classification on early fitness trajectories: for each trial, we compute the cumulative median of best-per-generation fitness and test it against a threshold derived by maximizing the F1 score on an initial 90-trial dataset. The resulting rule (generation G* = 3, threshold T* = 0.140) achieves F1 = 0.872 on 180 independent validation trials, retaining over 90% of successful trials while cutting computational cost by 41.6%. Compared to Hyperband, our domain-specific rule is 64% more efficient with higher mean fitness, though Hyperband occasionally discovers higher peak solutions. On a converged search population the rule becomes too aggressive (recall 31.1%), motivating adaptive thresholds. The specific thresholds are ES-HyperNEAT-specific, but the methodology, deriving stopping criteria from fitness dynamics classification, is applicable to other evolutionary algorithms with stagnation-prone hyperparameter spaces.
Comments8 pages, 5 figures, 9 tables. Accepted version of a paper published at the 2026 IEEE Congress on Evolutionary Computation (CEC), part of the IEEE World Congress on Computational Intelligence (WCCI 2026), Maastricht, Netherlands, 21-26 June 2026. (c) 2026 IEEE
