INTRODUCTION TO THE TLS EXIN NEURON - Neural-based Orthogonal Data Fitting: The EXIN Neural Networks - page 134

Comparison GeTLS, partial SVD

2.5 × 10 5

2

1.5

1

0.5

0

− 0.5

0

0

500

200

400

600

n 800

1000

1000

m

Figure 4.13 Computational cost: partial SVD vs. TLS EXIN.

Ge TLSvs. inverse iteration (algorithm #2), ki = 2

12 × 10 4

10

8

6

4

2

0

0

500 200

400

600

800

1000 1000

n

m

Figure 4.14 Computational cost: inverse iteration method vs.TLS EXIN ( K i = 2 is assumed).