A variant of k-means that trims points beyond a Chebyshev-based distance threshold is claimed to reduce intra-cluster variance by up to 88.1% and boost F1 by 20.8%, yet the evaluation protocol appears to score the cleaned data only.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Anomaly Detection and Improvement of Clusters using Enhanced K-Means Algorithm
A variant of k-means that trims points beyond a Chebyshev-based distance threshold is claimed to reduce intra-cluster variance by up to 88.1% and boost F1 by 20.8%, yet the evaluation protocol appears to score the cleaned data only.