### Table 2. Experiments with Initial Vector Coding Problems.

"... In PAGE 9: ...1) solved, the initial guess was the solution found at the previous problem. The numerical results are presented in Table2 , where MS means the number of iter- ations performed by the trust region algorithm mentioned above and Time indicates the total time spent to solve (5.... ..."

### Table 7.1: Initialization Vectors Fault Primary Inputs

### Table 2.2: Initialization of the scrambler vector

### Table 4. Comparison of weight vector initialization methods.

2001

"... In PAGE 6: ...4 % B. Weight Vector Initialization Method Table4 shows the results on the accuracy comparison of two initialization methods under the same experimental environ- ments. In the case of the proposed method called the uniform distribution of initial weight vectors, the experimental results on both the learning data and the test data showed better per- formance than those of the initialization with random values which is regarded as a basic initialization method.... ..."

Cited by 21

### TABLE I. (Continued)

2001

Cited by 2

### Table 3.1 Performance of the various eigensolvers for computing nev eigenvalues 1D-problems of size n with initial vectors chosen randomly.

### TABLE 1 Shape stability with different initializations of vector quantization Feature Iss NormX NormY Shell Sector

2002

### TABLE 2 Shape stability with different initializations of vector quantization (2) Feature Size (Vox) (z, y, x) Cross-Corr

2002

### Table 2 Di erent initial conditions and corresponding matrices, vectors and networks

"... In PAGE 17: ... Which of these two cases happened can sometimes be seen in the vector ks(t). When looking at the example shown in Table2 , one can see a further problem that occurs when generating the network out of the matrices: It is not always possible to decide whether the entries in the diagonal of the matrix belong to the degradation rate, which is assumed to be a linear function in x and which we did not show in our networks, or if they describe an autocorrelation for the relating gene, which is described as piecewise linear. Here, the model still needs further development.... In PAGE 18: ... The underlying question is, what behaviour the class of di erential equations consisting of piecewise linear equations is able to show, independent of special parameters. Therefore, we start looking at a single cuboid that de nes a certain network structure as was shown in Table2 . We solve the corresponding system of di erential equations for an arbitrary set of parameters and apply this afterwards to a network with given parameters.... ..."