On expressiveness and uncertainty awareness in rule-based classification for data streams
- Submitting institution
-
The University of Reading
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 71000
- Type
- D - Journal article
- DOI
-
10.1016/j.neucom.2017.05.081
- Title of journal
- Neurocomputing
- Article number
- -
- First page
- 127
- Volume
- 265
- Issue
- -
- ISSN
- 0925-2312
- Open access status
- Compliant
- Month of publication
- -
- Year of publication
- 2017
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
4
- Research group(s)
-
9 - DSAI
- Citation count
- 9
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- This paper presents a novel Hoeffding Rules data stream classification algorithm to produce an expressive rule set that adapts to concept drift in real-time. In comparison to other data stream classifiers, Hoeffding Rules are able to explain how a decision is reached and the algorithm can decide to abstain from classifying a data instance in case of uncertainty. This feature is desirable in applications where incorrect classification may be very costly such as in medical applications or in network intrusion detection. The experimental analysis shows that Hoeffing Rules outperforms direct competitors in terms of accuracy loss band and execution time.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -