A neural network that learns matrix-valued group representations by design achieves near-perfect accuracy and length extrapolation on group prediction tasks, outperforming fixed-representation, MLP, LSTM, and Transformer baselines.
Sampling using su (n) gauge equivariant flows
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
other 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
other 1polarities
unclear 1representative citing papers
citing papers explorer
-
MatrixNet: Learning over symmetry groups using learned group representations
A neural network that learns matrix-valued group representations by design achieves near-perfect accuracy and length extrapolation on group prediction tasks, outperforming fixed-representation, MLP, LSTM, and Transformer baselines.