HyperParameter Optimization (HPO) aims at finding the best HyperParameters (HPs) of learning models, such as neural networks, in the fastest and most efficient way possible. Most recent HPO algorithms try to optimize HPs regardless of the model that obtained them, assuming that for different models, same HPs will produce very similar results. We break free from this paradigm and propose a new take on preexisting methods that we called Genealogical Population Based Training (GPBT). GPBT, via the shared histories of "genealogically"-related models, exploit the coupling of HPs and models in an efficient way. We experimentally demonstrate that our method cuts down by 2 to 3 times the computational cost required, generally allows a 1% accuracy improvement on computer vision tasks, and reduces the variance of the results by an order of magnitude, compared to the current algorithms. Our method is search-algorithm agnostic so that the inner search routine can be any search algorithm like TPE, GP, CMA or random search.
神经网络在现代机器学习领域占据主导地位,但它们的训练和成功仍然受到对模型架构、损失函数和优化算法等超参数的经验选择的敏感性的影响。在这项工作中,我们提出了\emph{Population Based Training (PBT)},一个简单的异步优化算法,有效地利用固定的计算预算,共同优化人口的模型和他们的超参数,以最大限度地提高性能。重要的是,PBT发现了一个超参数设置的时间表,而不是遵循通常的次优策略,即试图找到一个用于整个训练过程的固定集合。只需对典型的分布式超参数训练框架进行一个小的修改,我们的方法就可以对模型进行鲁棒和可靠的训练。我们证明了PBT在深度强化学习问题上的有效性,通过优化一套超参数,显示出更快的挂钟收敛和更高的代理最终性能。此外,我们还展示了同样的方法可以应用于机器翻译的监督学习,其中PBT用于直接最大化BLEU分数,以及生成对抗网络的训练,以最大化生成图像的初始分数。在所有情况下,PBT都会自动发现超参数计划和模型选择,从而实现稳定的训练和更好的最终性能。
Two-Time-Scale Learning Dynamics: A Population View of Neural Network Training
Maxwell Institute for Mathematical Sciences and School of Mathematical and Computer Sciences Heriot–Watt University, Edinburgh, UK(麦克斯韦数学科学研究所和赫里奥特-沃特大学数学与计算机科学学院,爱丁堡,英国);Department of Mathematics and Computer Science University of Ferrara, Italy(费拉拉大学数学与计算机科学系)
Reinforcement-Guided Hyper-Heuristic Hyperparameter Optimization for Fair and Explainable Spiking Neural Network-Based Financial Fraud Detection
organization= Department of Computer Science \& Engineering, Khulna University of Engineering \& Technology (KUET) , city= Khulna , postcode= 9203 , country= Bangladesh;organization= Thomas Lord Department of Computer Science, Viterbi School of Engineering, University of Southern California , city= Los Angeles , state= CA , postcode= 90089 , country= USA;organization= Department of Computer Science, American International University-Bangladesh (AIUB) , city= Dhaka , postcode= 1229 , country= Bangladesh
Fast Efficient Hyperparameter Tuning for Policy Gradients
University of Oxford(牛津大学)
策略梯度方法的性能对必须针对任何新应用程序进行调优的超参数设置很敏感。用于调整超参数的广泛使用的网格搜索方法是样本效率低且计算昂贵的。更先进的方法,如基于人口的训练,学习超参数的最佳时间表,而不是固定的设置,可以产生更好的结果,但也是样本效率低,计算成本高。在本文中,我们提出了Hyperparameter Optimization on the Fly(HOOF),这是一种无梯度算法,它只需要一次训练就可以自动调整直接通过梯度影响策略更新的超参数。其主要思想是使用现有的轨迹采样的策略梯度方法来优化一步改进目标,产生一个样本和计算效率高的算法,易于实现。我们在多个领域和算法上的实验结果表明,使用HOOF来学习这些超参数计划可以提高学习速度和性能。
A Linear Programming Enhanced Genetic Algorithm for Hyperparameter Tuning in Machine Learning
Brij Disa Centre for Data Science and Artificial Intelligence(布里吉·迪萨数据科学与人工智能中心);Indian Institute of Management Ahmedabad(印度管理学院艾哈迈达巴德分校);Department of Statistics and Data Science(统计与数据科学系);Indian Institute of Technology Kanpur(印度理工学院坎普尔分校)
OptABC: an Optimal Hyperparameter Tuning Approach for Machine Learning Algorithms
Knight Foundation School of Computing and Information Sciences, Florida International University(佛罗里达国际大学奈特基金会计算与信息科学学院);Sustainability, Optimization, and Learning for InterDependent networks (solid) laboratory, FIU(佛罗里达国际大学可持续性、优化与相互依赖网络学习实验室);Department of Computer Science, University of Georgia(佐治亚大学计算机科学系)
ULTHO: Ultra-Lightweight yet Efficient Hyperparameter Optimization in Deep Reinforcement Learning
Department of Computing, The Hong Kong Polytechnic University(香港理工大学计算机系);Ningbo Institute of Digital Twin, Eastern Institute of Technology(宁波数字孪生研究所);Ningbo Key Laboratory of Spatial Intelligence and Digital Derivative(宁波空间智能与数字衍生关键实验室)