A linear regression from pretrained neural-network image features into multidimensional-scaling similarity spaces beats a baseline and peaks at two target dimensions, while metric and nonmetric scaling produce nearly equivalent spaces on the NOUN object set.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Generalizing Psychological Similarity Spaces to Unseen Stimuli
A linear regression from pretrained neural-network image features into multidimensional-scaling similarity spaces beats a baseline and peaks at two target dimensions, while metric and nonmetric scaling produce nearly equivalent spaces on the NOUN object set.