Resource management and scalability of the XCSF learning classifier system

Resource management and scalability of the XCSF learning classifier system
复制标题

XCSF学习分类器系统的资源管理和可扩展性

DOI:
10.1016/j.tcs.2010.07.007
复制
发表时间:
2012
期刊:
Theor. Comput. Sci.
影响因子:
--
通讯作者:
Goldberg
Goldberg
中科院分区:
--
文献类型:
--
作者:
Stalph;Llorà;Goldberg

文献摘要

参考文献

被引文献

相似文献

多次证明,进化在线学习XCS分类器系统是一个鲁棒泛化的强化学习系统,它在数据挖掘应用中也产生了极具竞争力的结果。该系统的 XCSF 版本是一个实值函数逼近系统,它学习分段重叠局部线性模型来逼​​近迭代采样函数。虽然二进制域方面的理论表明 XCS 可以 PAC 学习一组稍微受限的 k-DNF 问题,但 XCSF 的理论仍然相当稀疏。本文采用 XCS 方面的理论并将其投影到实值 XCSF 域。对于给出适应度指导的一组函数,我们甚至表明 XCSF 根据种群大小进行最佳缩放,仅需要恒定的开销来确保进化过程可以局部优化进化结构。因此,我们为 XCSF 提供了有关可扩展性和资源管理的基础。此外,我们揭示了 XCSF 以及一般本地线性学习器的问题难度维度,展示了结构对齐(即 XCSF 解决方案表示与问题结构的对齐)如何能够将挑战性问题的复杂性降低几个数量级。
It has been shown many times that the evolutionary online learning XCS classifier system is a robustly generalizing reinforcement learning system, which also yields highly competitive results in data mining applications. The XCSF version of the system is a real-valued function approximation system, which learns piecewise overlapping local linear models to approximate an iteratively sampled function. While the theory on the binary domain side goes as far as showing that XCS can PAC learn a slightly restricted set of k-DNF problems, theory for XCSF is still rather sparse. This paper takes the theory from the XCS side and projects it onto the real-valued XCSF domain. For a set of functions, in which fitness guidance is given, we even show that XCSF scales optimally with respect to the population size, requiring only a constant overhead to ensure that the evolutionary process can locally optimize the evolving structures. Thus, we provide foundations concerning scalability and resource management for XCSF. Furthermore, we reveal dimensions of problem difficulty for XCSF — and local linear learners in general — showing how structural alignment, that is, alignment of XCSF’s solution representation to the problem structure, can reduce the complexity of challenging problems by orders of magnitude.
使用 XCS 进行函数逼近:超椭球条件、递归最小二乘法和压缩
DOI: 10.1109/tevc.2007.903551
发表时间: 2008
影响因子: 14.3
作者:
Martin Volker Butz;P. Lanzi;Stewart W. Wilson
通讯作者: Stewart W. Wilson
使用 XCSF 进行上下文相关预测和认知手臂控制
DOI: 10.1145/1389095.1389360
发表时间: 2008
期刊: Proceedings of the 11th Annual conference on Genetic and evolutionary computation
影响因子: --
作者:
Martin Volker Butz;O. Herbort
通讯作者: O. Herbort
使用基因表达编程的分类器条件
DOI: --
发表时间: 2008
期刊: International Workshop on Learning Classifier Systems
影响因子: --
作者:
Stewart W. Wilson
通讯作者: Stewart W. Wilson
DOI: --
发表时间: 2005
期刊:
影响因子: --
作者:
Christopher Stone;L. Bull
通讯作者: L. Bull
XCSF 的预测更新算法:RLS、卡尔曼滤波器和增益自适应
DOI: --
发表时间: 2006
期刊: Annual Conference on Genetic and Evolutionary Computation
影响因子: --
作者:
P. Lanzi;D. Loiacono;Stewart W. Wilson;D. Goldberg
通讯作者: D. Goldberg