An embedding plus bilinear attention plus permuted CNN architecture reports higher accuracy than the cited tree and deep baselines on several small scientific tabular datasets.
Here we test GCN only on data without FRPs
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
EAPCR: A Universal Feature Extractor for Scientific Data without Explicit Feature Relation Patterns
An embedding plus bilinear attention plus permuted CNN architecture reports higher accuracy than the cited tree and deep baselines on several small scientific tabular datasets.