arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2601.03166cs.LG
收藏

动态超参数重要性用于高效多目标优化

Dynamic Hyperparameter Importance for Efficient Multi-Objective Optimization

Institute of Artificial Intelligence (LUH | | AI)(人工智能研究所);Leibniz University Hannover(汉诺威莱比锡大学);L3S Research Center(L3S研究中心)

Daphne Theodorakopoulos, Marcel Wever, Marius Lindauer

AI总结:

本文提出了一种动态优化方法,通过动态调整超参数重要性来提高多目标优化的效率和效果。

摘要:

选择合适的机器学习模型是一个复杂的过程,可能依赖于多个目标,例如准确性、模型大小、公平性、推理时间或能耗。在实践中,这需要通过多目标优化(MOO)在多个经常相互竞争的目标之间进行权衡。然而,现有的MOO方法通常将所有超参数视为同等重要,忽略了超参数重要性(HPI)会根据目标之间的权衡而显著变化。我们提出了一种新的动态优化方法,在搜索过程中根据变化的目标权衡优先考虑最影响性的超参数,这加速了经验收敛并导致更好的解决方案。基于先前关于MOO后分析的HPI工作,我们现在将使用HyperSHAP计算的HPI整合到优化中。为此,我们利用MOO算法ParEGO自然产生的目标权重,通过固定不重要的超参数来适应配置空间,使搜索能够专注于重要的超参数。最终,我们通过PyMOO和YAHPO-Gym中的多样化任务验证了我们的方法。实验证明,与基线相比,在收敛速度和Pareto前沿质量方面有所改进。

英文摘要:

Choosing a suitable ML model is a complex task that can depend on several objectives, e.g., accuracy, model size, fairness, inference time, or energy consumption. In practice, this requires trading off multiple, often competing, objectives through multi-objective optimization (MOO). However, existing MOO methods typically treat all hyperparameters as equally important, overlooking that hyperparameter importance (HPI) can vary significantly depending on the trade-off between objectives. We propose a novel dynamic optimization approach that prioritizes the most influential hyperparameters based on varying objective trade-offs during the search process, which accelerates empirical convergence and leads to better solutions. Building on prior work on HPI for MOO post-analysis, we now integrate HPI, calculated with HyperSHAP, into the optimization. For this, we leverage the objective weightings naturally produced by the MOO algorithm ParEGO and adapt the configuration space by fixing the unimportant hyperparameters, allowing the search to focus on the important ones. Eventually, we validate our method with diverse tasks from PyMOO and YAHPO-Gym. Empirical results demonstrate improvements in convergence speed and Pareto front quality compared to baselines.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑