TVSCM constrains fully connected layers to two-valued symmetric circulant weight matrices, delivering over 80x parameter reduction with accuracy falling from 97.6% to 93.5% on MNIST and 93.1% on MIT-BIH arrhythmia data.
Solving a Low-Rank Factorization Model for Matrix Completion by a Nonlinear Successive Over-Relaxation Algorithm.Mathematical Programming Computation, 4(4):333–361, 2012
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
-
Two-Valued Symmetric Circulant Matrices: Applications in Deep Learning
TVSCM constrains fully connected layers to two-valued symmetric circulant weight matrices, delivering over 80x parameter reduction with accuracy falling from 97.6% to 93.5% on MNIST and 93.1% on MIT-BIH arrhythmia data.