计算智能新进展：数据分析方法的深度探讨

需积分: 9 76 浏览量更新于2024-07-18 1 收藏 13.19MB PDF 举报

"《计算智能研究》系列（Studies in Computational Intelligence, SCI）是一套致力于快速、高质量发布计算智能领域新进展和先进技术的系列丛书。该系列涵盖了工程、计算机科学、物理、生命科学等多个学科，深入探讨了计算智能的理论、应用和设计方法。内容涉及神经网络、连接主义系统、遗传算法、进化计算、人工智能、细胞自动机、自组织系统、软计算、模糊系统以及混合智能系统等多个子领域。本书《数据智能分析中的计算智能方法进展》是SCI系列的第738卷，由Adam Gawęda、Janusz Kacprzyk、Leszek Rutkowski和Gary G. Yen共同编著，特别献给Jacek Żurada教授。该书反映了当前在数据智能分析中计算智能方法的最新研究成果，强调了这些方法在工程实践、科学研究和理论探索中的重要性。通过短篇出版周期和全球发行，系列书籍能够迅速传播研究成果，为学术界和产业界提供了丰富的学习资源和实践经验。《计算智能研究》系列的特点在于其广泛的主题覆盖，不仅限于单一的技术，而是融合了多学科的方法论，旨在推动计算智能技术的创新和发展。无论是研究人员、工程师还是学生，都能在这一系列中找到与各自专业领域相关的深度学习、机器学习和其他智能计算技术的深入讨论和案例研究。对于想要了解和应用计算智能解决实际问题的读者来说，这是一本不可或缺的参考文献和学习资料。"

integration of rule-based systems, notably: high-level modeling techniques for rule

bases, inte gration architect ures for rule-based systems, and rule interoperability.

A human assisted and an automatic derivation of rules are discussed, and some

challenging common problems, notably the handling of large rules sets through

structuring, integration of rule-based components, as well as rule interoperability

issues, are discussed.

Krystian Łapa, Krzysztof Cpałka, and Leszek Rutkowski (“New Aspects of

Interpretability of Fuzzy Systems for Nonlinear Modeling”) discuss fuzzy systems

as a well suited tool for modeling nonlinear systems. The authors emphasize that

the fuzzy systems can be effectively and efﬁciently used if their structure and

structure parameters are properly chosen, and the rules are clear and interpretable.

A new algorithm for the automatic learning of fuzzy systems and new inter-

pretability criteria of fuzzy syst ems are proposed. The interpretability criteria are

related to all aspect s of those systems, not only their fuzzy sets and rules, and also

concern the choice and analysis of parameterized triangular norms, discretization

points and weights of importance from the rules. Such a compr ehensive solution is

novel. The proposed criteria are taken into account in the learning process which

proceeds using a new learning algorithm that combines the genetic algorithm and

the ﬁrework algorithms, which makes it possible to automatically choose not only

the parameters but also the structure of the system. The new approach is tested on

some relevant simulation problems of nonlinear modeling.

Krassimir T. Atanassov and Peter Vassilev discuss in their paper “On the

Intuitionistic Fuzzy Sets of n-th Type” the use of various extensions of the concept

of a fuzzy set introduced by Zadeh, notably some extensions along the line of

Atanassov’s intuitionistic fuzzy set that makes it possible not only to express

imprecision of information but a very important problem related to the fact that the

human beings tend to use in their everyday discourse, judgments, reasoning, etc.,

aspects for and against. The author clariﬁes some misconceptions and introduces a

uniﬁed framework for such approaches.

In Part IV, “Intelligent Technologies in Decision Making, Optimization and

Control,” the ﬁrst paper by Jacek Mańdziuk (“MCTS/UCT in solving real-life

problems”) deals with the Monte Carlo Tree Search (MCTS) supported by the Upper

Conﬁdence Bounds Applied to Trees (UCT) method, i.e., the so-called MCTS/UCT

which is one of the stat e-of-the-art techniques in the game-playing domain. In

particular, it is emphasized the spectacular success of this method (combined with

the use of deep neural networks trained with the reinforcement learning algorithm)

in the game of Go. The author summarizes his works and experience in the appli-

cation of MCTS/UCT to domains other than games, with a particular emphasis on

hard real-life problems with a large degree of uncertainty due to the existence of

some stochastic factors in their deﬁnition, exempliﬁed by the Capacitated Vehicle

Routing Problem with Trafﬁc Jams, and the Risk-Aware Project Scheduling

Problem. It is shown how MCTS/UCT is a viable method in these two domains,

notably due its ability to effectively and efﬁciently deal with uncertainty by online

adaptation of the core MCTS simulations to the current situation.

Preface xv

Miłosz Kadziński, Micha ł K. Tomczyk, and Roman Słowiński (“Interactive

Cone Contraction for Evolutionary Multiple Objective Optimization”) present a

new interactive evolutionary algorithm for Multiple Objective Optimization

(MOO) which combines the NSGA-II method with a cone contraction method. The

new approach requires the Decision Maker (DM) to provide the preference infor-

mation as a reference point and pairwise comparisons of solutions from a current

population. This information is represented using a compatible Achievement

Scalarizing Function (ASF) which is used to guide the evolutionary search toward

the most preferred region of the Pareto front. The proposed algorithm is tested on a

set of benchmark problems, and the results show its quick convergence to the DM’s

most preferred region. Moreover, it also indicated the advantage of the new algo-

rithm of the well-known NEMO-0, in particular when the DM provides a richer

preference information composed of a greater number of pairwise comparisons of

solutions.

Oscar Castillo, Carlos Soto, and Fevrier Valdez (“A Review of Fuzzy and

Mathematic Methods for Dynamic Parameter Adaptation in the Fireﬂy Algorithm”)

are concerned with some issues related to the design and use of the ﬁreﬂy algorithm,

a well-known meta-heuristic. The authors concentrate on the choice of parameters

of the ﬁreﬂy algorithm, its analysis, and dynamic adjustments. Some relevant tra-

ditional and fuzzy logic-based approaches are analyzed and numerically compared.

In Part V, “Applications of Intelligent Technologies,” in the ﬁrst paper by

Adam E. Gawęda and Michael E. Brier (“Computational Intelligence Methods in

Personalized Pharmacotherapy”), the authors are concerned with a pharmacologic

therapy of chronic diseases that remains a big challenge to physicians, notably

because individual dose-response characteristics of patients may vary signiﬁcantly

across patient populations, and—due to a chronic nature of the process—they may

change over time within individual patients as well. Current state-of-the-art pro-

tocols for dose adjustment of pharmacologic agents rely heavily on data from the

drug approval process and a physician’s expertise but they do not fuzzy utilize the

wealth of knowledge hidden in patient data collected during his or her treatment.

The authors review the application of two computational intelligence methods: the

artiﬁcial neural networks and fuzzy sets theory, to personalized pharmaco logic

treatment of a chronic condition using patient data. As an example, the authors use

data on patients with anemia and renal failure.

Zdzisław Kowalczuk and Michał Czubenko (“Embodying Intelligence in

Autonomous and Robotic Systems with the Use of Cognitive Psychology and

Motivation Theories”) discuss a coherent anthropological approach for the control

of autonomous robots or agents. This modern approach is based on an appropriate

modeling of the human mind using the available psychological knowledge. One

of the main reasons that have inspired the authors is the lack of available

and effective top-down approac hes resulting from the some known results from the

area of autonomous robotics. On the other hand, a system for a comprehensive and

effective and efﬁcient modeling of human psychology for the purpose of con-

structing autono mous systems is lacking. The authors review the recent progress in

the understanding of the mecha nisms of cognitive computations underlying

xvi Preface

decision-making and existing challenges, notably those founded on cognitive ideas

such as LIDA, CLARION, SOAR, MANIC, DUAL, and OpenCog. In particula r,

the idea of an Intelligent System of Decision-making (ISD) is emphasized that is

based on the results of cognitive psychology (using the aspect of “information

path”), motivation theory (where the needs and emotions serve as the main drives,

or motivations, in the mechanism of governing autonomous systems), and several

other detailed theories, which concern memory, categorization, perception, and

decision-making. In the ISD system, in particular, an xEmotion subsystem is

focused on that covers the psychological theories on emotions, including the

appraisal, evolutionary, and somatic theories.

Krystian Łapa and Krzysztof Cpałka (“Evolutionary Approach for Automatic

Design of PID Controllers”) present a new approach to an automatic design of the

well-known and widely used PID controllers. It is based on a meta-heuristic hybrid

algorithm which combines the genetic algorithm and the imperialist one. The main

characteristic of the proposed approach is its capability to design the structure of the

controller and the structure of its parameters. This eliminates the need for a

trial-and-error process during the design of the controller structure. Moreover, in the

proposed approach, various control criteria can be reﬂected.

Marcin Zalasiński, Krzysztof Cpałka, and Leszek Rutkowski (“Fuzzy-genetic

Approach to Identity Veriﬁcati on Using a Handwritten Signature”) discuss a rele-

vant biometric problem of the veriﬁcation of the dynamic signature. There are many

methods for the signature veriﬁcation using dynamics of the signing process often

based on the so-called global features. In this paper, a new approach to the signature

veriﬁcation using global featu res is proposed. Basically, it involves the classiﬁca-

tion of the signature which is performed using a fuzzy-genetic system; the selec tion

of an individual set of features for each signer which uses a genetic algorithm with

an appropriately desig ned evaluation function and works without access to the

signatures called skilled forgeries; and the determination of weights of importance

for evolutionarily selected features which are taken into account in the classiﬁcation

process. The main advantages of this new approach is that the feature selection via a

fuzzy-genetic systems works with access to the signatures called skilled forgeries,

and also that the proposed classiﬁer can do without machine learning with respect to

its work interpretation and possibility of an analytical determination of its parameters.

Simulation results for the BioSecure signature database, distributed by the BioSecure

Association, are performed and conﬁrm the above mentioned good results.

S. Piasecki, R. Szmurlo, J. Rabkowski, and M.P. Kaźmierkowski (“A Meth od of

Design and Optimization for SiC-based Grid-connected AC-DC Converters”)

present a method of design and op timization for three-phase AC-DC converters.

The main idea of presented work is to provide a tool which supports the design

process and helps to achieve the main desired properties: efﬁciency, volume,

weight, and cost. The proposed design method is described with a special attention

paid to calculations regarding the power section of the converter. The authors

concentrate on the new technology of SiC power devices. The method is illustrated

on three SiC-based laboratory models rated at 10 and 20 kVA, respectively.

Preface xvii

剩余416页未读，继续阅读

qq_21583515

粉丝: 0
资源: 16

计算智能新进展：数据分析方法的深度探讨

最新资源