深度学习新视角：基于粒计算的机器学习在大数据处理中的进展

需积分: 0 44 浏览量更新于2024-07-18 收藏 2.1MB PDF 举报

《基于粒计算的机器学习：大数据处理方法》是一本探讨大数据、机器学习与粒计算之间关系的专业著作。作者汉·刘和米哈埃拉·科切亚，分别来自卡迪夫大学和朴茨茅斯大学计算机科学学院，将焦点放在如何利用粒计算的概念提升机器学习的深度，使之从浅层学习向深层学习（广义上的深度学习）转变，以更好地应对大数据时代日益复杂的学习任务。在大数据的背景下，传统的机器学习被看作是单粒度的学习，其处理能力受到限制。作者提出将机器学习方法从单一粒度扩展到多粒度，通过以下几种关键转变实现这一目标： 1. 从监督学习到半监督学习：通过引入部分标记数据，提升模型对未标记数据的泛化能力，增强学习的效率。 2. 从启发式学习到半启发式学习：结合规则和数据驱动的方法，使模型更具适应性和灵活性。 3. 从单任务学习到多任务学习：通过共享信息和知识，让模型能够同时解决多个相关问题，提高整体性能。 4. 从判别性学习到生成性学习：改变学习策略，使模型不仅能预测结果，还能理解和生成数据的潜在结构。 5. 从随机数据划分到半随机数据划分：通过更有序的数据组织，提高学习算法的稳定性和准确性。此外，书中还深入研究了如何在多粒度学习环境下，对属性值对进行深度评估，以诱导出高质量的规则，从而促进知识的高效提取和理解。通过这些方法，作者旨在构建出能够有效处理大规模数据的新型机器学习技术，以适应不断增长的信息挑战。本书不仅提供了理论框架，也为实践者提供了一套实用工具，帮助他们在实际工作中提升大数据处理能力。

8 1 Introduction

In Chap. 3, based on the issues of supervised learning identiﬁed in Chap. 2,we

present the proposal of semi-supervised learning towards addressing the issues espe-

cially in the context of big data processing. In particular, we describe the concepts

of semi-supervised learning and illustrate the proposed framework of learning in the

context of granular computing.

In Chap. 4, based on the issues of heuristic learning identiﬁed in Chap. 2,we

present the proposal of semi-heuristic learning, which is naturally and biologically

inspired, towards addressing the issues especially in the context of big data process-

ing. In particular, we describe the concepts of semi-heuristic learning and illustrate

the nature and biology inspired framework of learning in the context of granular

computing.

In Chap. 5, based on the issues of single-task learning and discriminative learning,

we present the proposal of fuzzy classiﬁcation through generative multi-task learn-

ing, towards addressing the issues especially in the context of big data processing.

In particular, we describe the concepts of both multi-task learning and generative

learning, and illustrate the framework of fuzzy generative multi-task learning.

In Chap. 6, based on the issues of random data partitioning identiﬁed in Chap. 2,

we present the proposal of semi-random data partitioning. In particular, we describe

the concepts of data partitioning and illustrate the multi-granularity framework of

semi-random data partitioning.

In Chap. 7, based on the issues of rule learning identiﬁed in machine learning lit-

erature, we presents the proposal of multi-granularity rule learning. In particular, we

describe the concepts of rule learning and illustrate the multi-granularity framework

towards in-depth learning of rules.

In Chap. 8, we present several case studies of big data by using biomedical data

and sentiment data, respectively, in order to show how big data processing can be

advanced through the shift from traditional machine learning to granular computing

based machine learning.

In Chap. 9, we will stress the theoretical signiﬁcance, practical importance,

methodological impact and philosophical aspects of granular computing based

machine learning, and several further directions are suggested towards machine learn-

ing advances to ﬁt the needs of modern industries.

References

1. Liu, H., A. Gegov, and M. Cocea. 2016. Rule based systems for big data: a machine learning

approach. Switzerland: Springer.

2. Pedrycz, W., and S.-M. Chen. 2015. Information granularity, big data, and computational

intelligence. Heidelberg: Springer.

3. Pedrycz, W. , and S.M. Chen. 2017. Data science and big data: an environment of computational

intelligence. Heidelberg: Springer.

4. P. Levine. Machine learning + big data. http://a16z.com/2015/01/22/machine-learning-big-

data/

References 9

5. Liu, H., A. Gegov, and M. Cocea. 2017. Uniﬁed framework for control of machine learning

tasks towards effective and efﬁcient processing of big data. In Data Science and Big Data: An

Environment of Computational Intelligence, 123–140. Switzerland: Springer.

6. Wu, X., X. Zhu, G.-Q. Wu, and W. Ding. 2014. Data mining with big data. IEEE Transactions

on Knowledge and Data Engineering 26 (1): 97–107.

7. Siddique, N., and H. Adeli. 2013. Computational intelligence: synergies of fuzzy logic, neural

networks and evolutionary computing. New Jersey: Wiley .

8. Rutkowski, L. 2008. Computational intelligence: methods and techniques. Heidelberg:

Springer.

9. J. Worrell. 2014. Computational learning theory: 2014–2015. https://www.cs.ox.ac.uk/

teaching/courses/2014-2015/clt/

10. H. Liu, M. Cocea, and A. Gegov. 2016. Interpretability of computational models for sentiment

analysis. In Sentiment Analysis and Ontology Engineering: An Environment of Computational

Intelligence, W. Pedrycz and S.-M. Chen, eds., vol. 639, 199–220.

11. Tan, P.-N., M. Steinbach, and V. Kumar. 2006. Introduction to Data Mining. New Jersey:

Pearson Education.

12. Mitchell, T. 1997. Machine Learning. New York: McGraw Hill.

13. H. Liu, A. Gegov, and M. Cocea. 2015. Network based rule representation for knowledge dis-

covery and predictive modelling. In IEEE International Conference on Fuzzy Systems,Istanbul,

Turkey, 2–5 August 2015, 1–8.

14. H. Liu, A. Gegov, and F. Stahl. 2014. Categorization and construction of rule based systems.

In 15th International Conference on Engineering Applications of Neural Networks,Soﬁa,

Bulgaria, 5–7 September 2014, 183–194.

15. Quinlan, R.J. 1986. Induction of decision trees. Machine Learning 1 (1): 81–106.

16. Furnkranz, J. 1999. Separate-and-conquer rule learning. Artiﬁcial Intelligence Review 13: 3–54.

17. J. Zhang. 1992. Selecting typical instances in instance-based l earning. In Proceedings of the

Ninth International Workshop on Machine Learning, Aberdeen, United Kingdom, 1–3 July

1992, 470–479.

18. Hazewinkel, M. 2001. Encyclopedia of Mathematics. London: Springer.

19. I. Rish. 2001. An empirical study of the naive bayes classiﬁer. IJCAI 2001 Workshop On

Empirical Methods In Artiﬁcial Intelligence, 3 (22), 41–46.

20. Breiman, L. 1996. Bagging predictors. Machine Learning 24 (2): 123–140.

21. Y. Freund and R. E. Schapire. 1996. Experiments with a new boosting algorithm. In Machine

Learning: Proceedings of the Thirteenth International Conference, Bari, Italy, 3–6 July 1996,

148–156.

22. Y. Yao. 2005. Perspectives of granular computing. In Proceedings of 2005 IEEE International

Conference on Granular Computing, Beijing, China, 25–27 July 2005, 85–90.

23. Merriam-Webster. 2016. Merriam-websters dictionary. http://www.merriam-webster.com/

24. Liu, H., A. Gegov, and M. Cocea. 2016. Rule based systems: A granular computing perspective.

Granular Computing 1 (4): 259–274.

25. Zadeh, L. 2015. Fuzzy logic: A personal perspective. Fuzzy Sets and Systems 281: 4–20.

26. Pedrycz, W. 2011. Information granules and their use in schemes of knowledge management.

Scientia Iranica 18 (3): 602–610.

27. H. Liu and M. Cocea. Fuzzy information granulation towards interpretable sentiment analysis.

Granular Computing, 3 (1), In press.

28. Pedrycz, W., and S.-M. Chen. 2011. Granular computing and intelligent systems: design with

information granules of higher order and higher type. Heidelberg: Springer.

29. Pedrycz, W., and S.-M. Chen. 2015. Granular computing and decision-making: interactive

and iterative approaches. Heidelberg: Springer.

30. Wilke, G., and E. Portmann. 2016. Granular computing a s a basis of humandata interaction: a

cognitive cities use case. Granular Computing 1 (3): 181–197.

31. Skowron, A., A. Jankowski, and S. Dutta. 2016. Interactive granular computing.

Granular

Computing 1 (2): 95–113.

12 2 Traditional Machine Learning

2.2 Heuristic Learning

Heuristic learning means learning with a speciﬁc strategy, which is based on statistical

heuristics, such as Bayes Theorem (Eq. 2.1) and distance functions (Eqs. 2.2 and 2.3).

P(Y |X ) =

P(Y )P(X|Y )

P(X)

(2.1)

where X and Y are two events:

• P(X) is read as the probability that event X occurs to be used as evidence sup-

porting event Y .

• P(Y ) is read as the prior probability that event Y occurs on its own.

• P(Y |X ) is read as the posterior probability that event Y occurs given that event X

truly occurs.

• P(X|Y ) is read as conditional probability that event X occurs subject to that event

Y must occur.

D(x, y) =









i=1

− y

)

(2.2)

where n is the dimensionality (the number of attributes) and i is the index of the

attribute. For example, if the data is in two dimensions, then Eq. 2.2 can be rewritten

as Eq. 2.3 as follows:

D(x, y) =



− y

)

+ (x

− y

)

(2.3)

Bayes’ Theorem is essentially used in Bayesian learning, and a popular algorithm

of this type of learning is Naive Bayes (NB), as mentioned in Chap. 1. The learning

outcome of this algorithm is ﬁnding the class that has the highest posterior probability

given all input attributes with their values as the joint condition. The learning strategy

of this algorithm is in the following procedure based on Bayes’ theorem:

Step 1: Calculating the posterior probability of each class given each attribute with

its value P(y = c

= v

) on the basis of training instances, where y is the class

attribute, c

is the k

class label of y, x is the input attribute, i is the index of attribute

x and v

is the j

value of x

Step 2: Calculating the posterior probability of a class:



i=0

P(y = c|x

= v),

where i is the index of attribute x and n is the number of attributes.

Step 3: Assigning the test instance the class that has the highest posterior probability

on the basis of all the attribute values.

An example shown in Table 2.1 is used for illustrating the procedure of the NB

algorithm:

剩余116页未读，继续阅读

wang1062807258

粉丝: 13
资源: 272

深度学习新视角：基于粒计算的机器学习在大数据处理中的进展

wx494社区门诊管理系统小程序-php+vue+uniapp.zip（可运行源码+sql文件+文档）

HTML+CSS+JS+JQ+Bootstrap的家具风格趋势展示响应式网页.7z

高分项目，基于Python+OpenCV的实时疲劳驾驶检测系统，内含源码+演示视频+部署教程

Granular Computing Based Machine Learning: A Big Data Processing Approach

Granular Computing Based Machine Learning A Big Data Processing 无水印原版pdf

Dominance-based rough set approach as a paradigm of knowledge discovery and granular computing (2010年)

granular computing

Hyperspherical granular computing classification algorithm based on fuzzy lattices

PyPI 官网下载 | granular-moonshot-0.3.0.tar.gz

Python库 | django-granular-access-0.1.tar.gz

最新资源