### Table 3. The local balance equations are:

in Form Bounds

2000

"... In PAGE 16: ...State No State No State No State 0 (C1;C1;T1) 3 (C1;C4;T3) 6 (C3;C1;T1) 9 (C3;C4;T3) 1 (C1;C2;T2) 4 (C2;C1;T2) 7 (C3;C2;T2) 10 (C4;C1;T3) 2 (C1;C3;T1) 5 (C2;C3;T2) 8 (C3;C3;T1) 11 (C4;C3;T3) Table3 : The states of the Markov chain of the TSS/TSSO systems 4.2.... In PAGE 19: ... Notice that this choice preserves the mean only. The corresponding Markov chain has 8 states and, if they are numbered ac- cording to the schema in Table3 , the local balance equations are: 1 = R13 0; 2=R24 0; 4=R13 0; 5=R13R24 0; 6 = R24 0; 7=R13R24 0; 8=R2 24 0; where R13 = R1=r3 and R24 = r2=r4. The probability 0 is equal to 1=dLow,where dLow =1+2R13 +2R24 +2R13R24 + R2 24:... ..."

### Table 1. Matrix and set of linear equations corresponding to the

1995

Cited by 1

### Table 2 Comparison of Models Based on Equation (3.1)

2001

Cited by 6

### Table 2: Energy preservation with Haar DWT for foetal ecg

2003

"... In PAGE 12: ... In general DFT will work better for signals with strong regular global patterns, while the multi-resolution property of the DWT makes it suitable for signals with locally occuring patterns. In Table2 you can see the results for the foetal ecg example compressed with the Haar DWT. For k 2 f3; 7; 15; 31; 63; 127; 255g 1 between 9% and 48% more energy of the original signal is preserved by the proposed method (see 1the rst three values of k represent a dimensionality typically used for indexing... In PAGE 15: ... 1 2 3 4 5 6 7 8 9 10 0 10 20 30 40 50 60 70 80 90 100 log2(k+1) preserved energy adapk eachk bestk firstk Figure 3: Energy preservation with Haar DWT for sp500 The results for all time series data sets and the Haar DWT are listed in Table 3. The percentage di erences listed correspond to the bold column of Table2 , that is the additional energy preserved by bestk over rstk . The col- umn with 1023 coe cients was omitted because using all coe cients always preserves all the energy.... ..."

Cited by 13

### Table 1: Computing time for one Verlet step (Intel Pen- tium 4, 2.8 GHz). The number of mass points is given for each model. Further, nD, nA, nV denote the num- bers of considered distance-preserving, area-preserving, and volume-preserving forces per integration step, re- spectively. In case of nA =0, the object surface is not considered in the deformation model (see Sec. 3.3).

2004

Cited by 23

### Table 1: Computing time for one Verlet step (Intel Pen- tium 4, 2.8 GHz). The number of mass points is given for each model. Further, nD, nA, nV denote the num- bers of considered distance-preserving, area-preserving, and volume-preserving forces per integration step, re- spectively. In case of nA =0, the object surface is not considered in the deformation model (see Sec. 3.3).

2004

Cited by 23

### Table 1. Times for deforming a body with a different number of triangles on the surface Number of

"... In PAGE 4: ...he quality of fit. There were a few cases like this. In order to check the algorithm complexity, we measured its speed for a different numbers of triangles on the surface. Results are given in Table1 . The times were measured using the profiling feature of Microsoft Visual C++ 6 and they are total times for computing the forces, integration of equations and rendering the image.... ..."

### Table 1 Deformability Index (DI) for Human Activities Using Motion Capture Data Activity DI Activity DI

"... In PAGE 17: ... We used the method described in Section 4 to the trajectories of these points to compute the DI for each of these sequences. These values are shown in Table1 for various activities. Please note that the DI is used to estimate the number of basis... In PAGE 18: ...shapes needed for 3D deformable object modeling, not for activity recognition. From Table1 , a number of interesting observations can be made. For the walk sequences, the DI is between 5 and 6.... In PAGE 19: ... 4(b). The row and column numbers correspond to the numbers in Table1 for 1-16, while 17 and 18 correspond to sitting and walking, where the training and test data are from two difierent viewing directions. For the moment, consider the upper 13 x 13 block of this matrix.... In PAGE 20: ... (b): The similarity matrix for the various activities, including ones with difierent viewing directions. The numbers correspond to the numbers in Table1 for 1-16. 17 and 18 correspond to sitting and walking, where the training and test data are from two difierent viewing directions.... In PAGE 20: ... In order to further show the efiectiveness of this approach, we used the ob- tained similarity matrix to analyze the recognition rates for difierent clusters of activities, We applied difierent thresholds on the matrix and calculated the recall and precision values for each cluster. The flrst cluster contains the walking sequences along with jogging and blind walk (activities 1-5,11, and 12 in Table1 ).... In PAGE 20: ...2 in Table 1). Fig. 5(a) shows the recall vs. precision values for this activity cluster, we can see from the flgure that we are able to identify 90% of these activities with a precision up to 90%. The second cluster consists of three sit- ting sequences (activities 6-8 in Table1 ), and the third cluster consists of the brooming sequences (activities 9 and 10 in Table 1). For both of these clusters the similarity values were quite separated to the extent that we were able to fully separate the positive and negative examples.... In PAGE 20: ...2 in Table 1). Fig. 5(a) shows the recall vs. precision values for this activity cluster, we can see from the flgure that we are able to identify 90% of these activities with a precision up to 90%. The second cluster consists of three sit- ting sequences (activities 6-8 in Table 1), and the third cluster consists of the brooming sequences (activities 9 and 10 in Table1 ). For both of these clusters the similarity values were quite separated to the extent that we were able to fully separate the positive and negative examples.... In PAGE 21: ...ig. 5. The recall vs. precision rates for the detection of three difierent clusters of activities. (a) Walking activities (activities 1-5,11, and 12 in Table1 ) (b) Sitting activities (activities 6-9 in Table 1 ) (c) Brooming activities (activities 9 and 10 in Table 1) image plane, two for jogging in a circle and one for brooming in a circle. We considered a portion of these sequences where the stick flgure is not parallel to the camera.... In PAGE 21: ...ig. 5. The recall vs. precision rates for the detection of three difierent clusters of activities. (a) Walking activities (activities 1-5,11, and 12 in Table 1) (b) Sitting activities (activities 6-9 in Table1 ) (c) Brooming activities (activities 9 and 10 in Table 1) image plane, two for jogging in a circle and one for brooming in a circle. We considered a portion of these sequences where the stick flgure is not parallel to the camera.... ..."

### Table 2: Comparison of three algorithms. The domain used in Table 1 has been deformed and the nite element discretization is now irregular. Iteration count and the corresponding operator condition number estimates are given.

1999

"... In PAGE 13: ... Iteration count and the corresponding operator condition number estimates are given. In Table2 , we consider the same methods as in the rst table, but now on a nite element problem where the elements has been signi cantly distorted. The resulting nite element mesh is now irregular.... ..."

Cited by 2

### Table 1 Frequency Corresponding to Each Observation in Equation 3.2

in and

1997

"... In PAGE 9: ...The relationship between the rows of A (the observations in Equation 3.2) and the corresponding frequencies is summarized in Table1 . The second and third rows of A are the coefficients in the finite sine and cosine transforms at the lowest non-zero frequency, (1/T); the fourth and fifth rows are the coefficients for the transforms at the next frequency, (2/T), etc.... In PAGE 11: ... Testing for Parameter Stability in the Frequency Domain Since the real-valued spectral regression model (Equation 3.2) is an ordinary regression equation the only difference being that its T observations correspond to the frequencies given in Table1 rather than to time periods the stability of the Y - X relationship across frequencies t tk can be tested using established methods for assessing the stability of regression coefficients. A number of such methods exist, including Chow (1960), Farley, Hinich, and McGuire (1975), Brown, Durbin, and Evans (1975), Garbade (1977), LaMotte and McWhorter (1978), Ashley (1984), and Watson and Engle (1985).... ..."