Shared-score quaternion self-attention reduces score multiplications by 75% and softmax operations from four to one while proving equivalence to component-wise attention under quaternion linear projections.
Learning to rotate: Quaternion transformer for complicated periodical time series forecasting
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
-
Quaternion Self-Attention with Shared Scores
Shared-score quaternion self-attention reduces score multiplications by 75% and softmax operations from four to one while proving equivalence to component-wise attention under quaternion linear projections.