### Table 5. Evaluation of algorithm AMD on a set of obfuscated variants of B[e]agle.Y. For com- parison, we include the detection rate of McAfee VirusScan.

2005

"... In PAGE 14: ... To test the limits of our imple- mentation, one of the math-op insertion transformations actually replaced single instructions of the form x = x + const with equiv- alent sequences. The results are tabulated in Table5 : our tool catches all obfuscations with the exception of the math-op re- placement transformation; in such a case, since the algorithm seeks to exactly match a template instruction of the form x = x + const, any equivalent instruction sequences are not detected.... ..."

Cited by 50

### Table 2. Square-error cost metric for various test images (error-diffusion initial image).

"... In PAGE 8: ...eeded to define efficient and perceptually meaningful cost functions (i.e., in our case filters) and/or better initial binary images to be used in the context of a DLM halftoning problems. As an example, Table2 illustrates cost metric values when the error diffusion images are used as initial images for the grid algorithm. While the numbers clearly show that our algorithm reduces the overall cost metric with respect to the initial condition {Floyd-Steinberg error diffusion is a one-pass non-iterative algorithm.... ..."

### TABLE 1 Mean percentages of errors to target stimuli presented at the four distances from the observer on the left side and on the right side in Experiment 1

2003

Cited by 1

### TABLE 4 Mean percentages of errors to target stimuli presented on the left side and on the right side at the four distances from the observer in Experiment 2

2003

Cited by 1

### Table 5. Evaluation of algorithm AMD on a set of obfuscated variants of B[e]agle.Y. For comparison, we include the detection rate of McAfee VirusScan.

"... In PAGE 11: ... To test the limits of our implementation, one of the math-op insertion transformations actually replaced single instructions of the form x = x + const with equivalent sequences. The results are tabulated in Table5... ..."

### Table 5. Evaluation of algorithm AMD on a set of obfuscated variants of B[e]agle.Y. For comparison, we include the detection rate of McAfee VirusScan.

"... In PAGE 11: ... To test the limits of our implementation, one of the math-op insertion transformations actually replaced single instructions of the form x = x + const with equivalent sequences. The results are tabulated in Table5... ..."

### Table 5. Evaluation of algorithm AMD on a set of obfuscated variants of B[e]agle.Y. For comparison, we include the detection rate of McAfee VirusScan.

"... In PAGE 11: ... To test the limits of our implementation, one of the math-op insertion transformations actually replaced single instructions of the form x = x + const with equivalent sequences. The results are tabulated in Table5... ..."

### Table 1. A Subset of Input 3D Image Data Set Modality #Slices Form and Size Voxel (mm) Pathology

"... In PAGE 1: ...lices with elongated voxel sizes (e.g. voxel sizes of 1:1:20 (mm3)) and with only a subportion of the brain visible in the set of scans. Table1 shows the parameters of a few sample input image sets. Due to the fact that the voxels of some 3D images can be far from cubical, the algorithm described here is a pixel-based instead of voxel-based 3D image algorithm.... In PAGE 9: ... The angles of the symmetry axes chosen by a neuroradiologist (curve with dots) versus the angles, i = arg max(C maxi( j)), computed by our algorithm. This is dataset (3) of Table1 . The algorithm determined angle is 2 degrees, the mean of the expert chosen angle is 1.... In PAGE 11: ...Figure 15. The mid-sagittal plane automatically extracted from the MR data set ( Table1 , data set (4)). The detected yaw is 1.... ..."