Skip to main content

Table 1 Comparing performance of different algorithms on a subset of a non-small cell lung cancer dataset, with k=15

From: Optimization and expansion of non-negative matrix factorization

 SCD-MSELEE-MSELEE-MSE-1SCD-MKLLEE-MKL
MSE0.1550.15650.15570.15740.1579
MKL0.011410.011490.011450.011190.01122
Rel. tol.1.325e-050.00013810.0001296.452e-089.739e-05
Total epochs50005000500050005000
Time (Sec.)1.3051.358.45649.1741.11
  1. MSE = mean square error; MKL = mean KL divergence; Rel. tol. = relative tolerance. Elapsed time = actual running time. SCD-MSE = SCD algorithm with MSE loss and 50 inner iterations and LEE-MSE-1 = Lee’s algorithm with MSE loss and 1 inner iteration, i.e., the original multiplicative algorithm