### Table 6: Table of results with tx = 1:5 pixels, ty = 1:3 pixels and tz = 0 for computing u0, v0 and ty tx from Cu and C ? v. Then, we compute the angle from ty tx and the result obtained by minimization (26) with this initialization is the same as with the previous method de ned by (18) and (21). As a conclusion we observe that we obtain a much better precision on u0 and v0 but that the good precision on ty tx in the case of t = (tx; ty; 0) and w = (0; 0; wz) decrease very fast with introducing of other components in the motion. Furthermore, we easily verify that considering a weighted estimation which integrate the fact that foveal information is more reliable that peripheral information signi cantly improve the performances of the results.

### Table 2 shows the pair-wise correlations of the values. As can be seen from the cross-correlations in the table, the parameter settings did not all converge to the same values. Nevertheless, there were some cases in which the parameters were highly correlated. In particular the solutions found in runs Tr1a, and Tr2a are highly similar (this can be easily verified by eye in figure 3. A rather strong correlation is also observed between the solutions found in Tr1c and Tr2b, and those in Tr1b, and Tr2c. It is interesting that the correlated solutions were obtained using non-overlapping sets of performances. This is evidence that the solutions found are approximations of a single parameter setting that is valid for the performances in both training sets. In the case of the solutions of Tr1a and Tr2a, the approximated parameter setting may also have a more general validity, since both solutions have a low error number of annotations on the test set as well (see table 1).

"... In PAGE 10: ... Table2 . Cross-correlations of the parameter values that were optimized using two... ..."

### Table 1: Example of structures parent, code, f and z satisfying the constraints (R1)-(R5)

2004

"... In PAGE 7: ...Figure 4: Tree corresponding to parent given in table 1 The above theorem suggests that there is a possibility that the tree formed by parent is not the same as the tree generated by the code. For example, consider the value of the variables given in Table1 . It can be easily verified that these values satisfy the constraint set C.... ..."

Cited by 1

### Table 1: Example of structures parent, code, f and z satisfying the constraints (R1)-(R5)

"... In PAGE 5: ...Figure 4: Tree corresponding to parent given in Table 1 The above theorem suggests that there is a possibility that the tree formed by parent is not the same as the tree generated by the code. For example, consider the value of the variables given in Table1 . It can be easily verified that these values satisfy the constraint set C.... ..."

### Table 1: Generalized spoke vectors Girth Spoke vector

"... In PAGE 2: ... Corollary 1 follows easily by verifying the two conditions given in Proposition 1 for the new graph. Table1 provides some examples of these general spoke vectors for various girths. The results of applying Corollary 1 the spoke vectors given by Exoo in [16] are shown in Table 2.... ..."

### Table 3: Taxonomic Knowledge

1998

"... In PAGE 5: ... Again, before focusing on their technical details, let us give some illustrating examples. Let B = fA; B; Cg and let KB = TKB [ PKB, where TKB is given by Table3 , left side, and PKB is given by the conjunctive events A, B, and C in Table 3, right side, and by the bounds in Table 4. We easily verify that all KB in (f) to (k) are coherent and consistent.... In PAGE 5: ... Again, before focusing on their technical details, let us give some illustrating examples. Let B = fA; B; Cg and let KB = TKB [ PKB, where TKB is given by Table 3, left side, and PKB is given by the conjunctive events A, B, and C in Table3 , right side, and by the bounds in Table 4. We easily verify that all KB in (f) to (k) are coherent and consistent.... ..."

Cited by 5

### Table 3. K is much less than N (see text). bmin is fixed to be the size of the largest frame for each trace

"... In PAGE 6: ... However, we find on experimentation with several VBR-compressed MPEG-1 video frame-size traces4 that K is in practice much smaller than N. This fact is illus- trated in Table3 . It is easily verified that Algorithm has a maxfO(K2);O(N)g complexity.... ..."

### Table 2. For each zone i the table gives the innermost cylinder, the outermost cylinder, the number of cylinders CB4iB5, and the number of requests DB4iB5.

"... In PAGE 6: ... When applied to the example, HDF initially generates all 15 possible intervals, and determines the one with the small- est inter-request gap. Using Table2 , it is easily verified that interval CU2BN3CV enforces the smallest gap. Hence, I1 BP CU2BN3CV.... ..."

### Table 1. Filter coefficients of Daubechies wavelet of order 8.

2005

"... In PAGE 13: ...ection 2.2. We choose the compact orthonormal Daubechies wavelet of order 8 to construct the high-pass and low-pass filters. Coefficients of Daubechies wavelet of order 8 are listed in Table1 . One could easily verify that the transformation matrix A similarly constructed as in equation (5) by these coefficients satisfies condition (7) to provide an orthogonal transfor- mation.... ..."

### Table 1. Examples of matrices with structure.

"... In PAGE 3: ... Similarly to the Pick ma- trices, their structure is understood in the sense that their D2BE entries are defined by a much smaller number C7B4D2B5 of parameters. Table1 presents some examples. Table 1.... In PAGE 4: ... The number ABB4CAB5 in (7) is called the displacement rank of CA, and if the displacement rank is small then CA is said to have a displacement structure. The [easily verified] facts in Table 2 show that the matrices listed in Table1 all have displacement structure. Many applications, however, go beoynd the simplest ex- amples of Table 1, and give rise to the more general classes of matrices with displacement structure.... In PAGE 4: ... The [easily verified] facts in Table 2 show that the matrices listed in Table 1 all have displacement structure. Many applications, however, go beoynd the simplest ex- amples of Table1 , and give rise to the more general classes of matrices with displacement structure. These are matri- ces with small displacement rank [i.... ..."