Controlled comparison on synthetic data shows objective choice in multiobjective unsupervised feature selection creates strong biases, with PCA reconstruction loss yielding compact subsets whose downstream accuracy matches direct supervised optimization.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Objective-Induced Bias and Search Dynamics in Multiobjective Unsupervised Feature Selection
Controlled comparison on synthetic data shows objective choice in multiobjective unsupervised feature selection creates strong biases, with PCA reconstruction loss yielding compact subsets whose downstream accuracy matches direct supervised optimization.