Core Concepts
Discovering top-k contrast patterns for effective time series classification.
Abstract
Introduction:
Time series mining methods proposed by various researchers.
Need for effective pattern mining to classify time series accurately.
Order-Preserving Pattern Mining:
OPP mining introduced to represent relative orders in time series.
Limitations of OPPs in capturing differences between classes.
Contrast Pattern Mining:
Contrast patterns can highlight significant differences between classes.
Importance of contrast patterns for classification accuracy.
COPP-Miner Algorithm:
Composed of extreme point extraction, forward mining, and reverse mining.
Steps involved in forward mining and reverse mining.
Experimental Results:
Efficiency and effectiveness of COPP-Miner algorithm validated.
Top-k COPPs identified as valuable features for classification.
Conclusion:
Summary of the proposed algorithm and its benefits for time series classification.
Stats
최소 대조율 cmin은 0보다 크다.
COPP-Miner 알고리즘은 2길이 패턴부터 시작한다.
COPP-Miner은 EPE 알고리즘을 사용하여 지역 극값을 추출한다.
Quotes
"Contrast patterns can be used to improve classification performance and model interpretability."
"COPP-Miner algorithm efficiently discovers top-k contrast patterns for time series classification."