On synthetic data, PCA reconstruction loss yields compact feature subsets whose supervised test accuracy matches direct optimization, while silhouette score strongly biases toward trivial low-cardinality solutions.
A taxonomy of unsupervised feature selection methods including their pros, cons, and challenges.The Journal of Supercomputing, 80(16):24212–24240, November 2024
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Objective-Induced Bias and Search Dynamics in Multiobjective Unsupervised Feature Selection
On synthetic data, PCA reconstruction loss yields compact feature subsets whose supervised test accuracy matches direct optimization, while silhouette score strongly biases toward trivial low-cardinality solutions.