Artificial Intelligence
Hierarchical censored production rules (HCPRs) system
Data & Knowledge Engineering
The application of AdaBoost for distributed, scalable and on-line learning
KDD '99 Proceedings of the fifth ACM SIGKDD international conference on Knowledge discovery and data mining
Mining high-speed data streams
Proceedings of the sixth ACM SIGKDD international conference on Knowledge discovery and data mining
Mining time-changing data streams
Proceedings of the seventh ACM SIGKDD international conference on Knowledge discovery and data mining
Data Mining and Knowledge Discovery with Evolutionary Algorithms
Data Mining and Knowledge Discovery with Evolutionary Algorithms
Mining concept-drifting data streams using ensemble classifiers
Proceedings of the ninth ACM SIGKDD international conference on Knowledge discovery and data mining
On demand classification of data streams
Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining
ACM SIGMOD Record
A parallel genetic algorithm approach for automated discovery of censored production rules
AIAP'07 Proceedings of the 25th conference on Proceedings of the 25th IASTED International Multi-Conference: artificial intelligence and applications
Expert Systems with Applications: An International Journal
| Hi-index | 0.00 |
As distributed data streams are gaining importance in a growing number of applications, the centralized approach for data mining is inappropriate for the distributed and ubiquitous data mining environments. The conventional mining algorithms for data streams depend on computationally expensive update procedures to incorporate the changing patterns in the streaming data. Moreover, the most common knowledge structure learnt, in knowledge discovery, is standard Production Rules (PRs) in the form: If P Then D. However, PRs ignore exceptions as noise. These are not efficient for approximate reasoning and unable to exhibit variable precision logic in the reasoning process due to rigidity in their structure. This paper proposes an evolutionary approach for distributed mining of CPRs using cumulative learning scheme. Local classifiers consisting of PRs and CPRs are generated for the data streams at distributed sites and then a meta-classifier is produced by combining the local classifiers. A Censored Production Rule (CPR) is an extension of PR and is of the form, If P Then D Unless C, where C is the censor representing exception condition. 'If P Then D' part of a CPR holds frequently and the censor C holds rarely. By using a rule of this type we are free to ignore the exception conditions, when the resources needed to establish its presence, are tight. Thus 'If P Then D' part of the CPR expresses important information while the 'Unless C' part acts only as a switch that changes the polarity of D to ~D, whenever a censor evaluates true A Genetic Algorithm is designed with a fixed length chromosome encoding that allows variable length rules. Appropriate genetic operators are suggested for the specific encoding and a fitness function incorporating the constraints of CPRs is formulated. Experimental results are presented to demonstrate the effectivity of CPR for cumulative learning in data streams.