Applying conjugate gradient to the hypergraph ranking system cuts computation time by about 92% on a 1,292-image dataset while preserving F1, but the block randomized SVD approach is not rigorously derived.
These methods are block randomized SVD for matrix inversion and conjugate gradient for solving a set of linear equations both related to optimizing f given fixed hyperedge weights w
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SI 1years
2019 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Block Randomized Optimization for Adaptive Hypergraph Learning
Applying conjugate gradient to the hypergraph ranking system cuts computation time by about 92% on a 1,292-image dataset while preserving F1, but the block randomized SVD approach is not rigorously derived.