Tools
Sorted by:
Try your query at:
 |
 |
 |
 |
 |
 |
Results 11 - 20
of
22,422
Table 1: Versioning Policies
"... In PAGE 6: ...olicies for that directory. Files in a directory that does not have a .version file are not versioned. A full list of the policies that may be specified is presented in Table1 . Each of these policies is detailed below.... ..."
Table 5. Improvement in top-30 performance provided by the window heu- ristic for all methods. The better score between the basic method and its windowed version is in bold. Full statistics are provided in the supplement.
2007
"... In PAGE 4: ... These relationships are not shown for the PPI data set, as the only significant differences on it between methods at the AUC1 level involve comparisons with the worst performing method. The top-30 improvement provided by using the window heuristic on each method on each data set is given in Table5 . Below we describe our main findings in detail.... In PAGE 5: ... Our heuristic for exploiting conserva- tion scores within a sequence window around the residue of interest can be applied to any scoring method that produces independent column scores. Using a window of size seven (three residues on eit- her side of the current residue), all methods improve on each of the three data sets ( Table5 ), as judged by top-30. The results are similar for the other statistics (see Supplementary material for all perfor- mance statistics for windowed approaches).... In PAGE 5: ... Note that the window method improves predictions for all types of functional sites, not just those with low conservation. In fact, as Table5 shows, the improve- ment is greater for sites with high conservation (catalytic residues and residues near ligands) than for sites in protein interfaces . Figure 3 shows the improvement on the CSA data set for SE and JSD when our window approach is used.... ..."
Table 2: Distortion Measurements age of 0.5 bits per pixel. Because the dimen- sionality of the output array was lower than that of the data, P lt; N, the code book did not achieve sorted order and the gradient search version of the Kohonen algorithm exhibited convergence problems. The results presented here are for only the full search version of the Kohonen algorithm and the K-Means al- gorithm. The following constants were used for training:
"... In PAGE 11: ... Figures 15{17 show the original image, and those coded with the K-Means and Koho- nen algorithms. Table2 compares the SNR for the K-Means and Kohonen algorithms. Keep in mind that these are distortion measure- ments for the training set only; there is no separate test set.... ..."
Table 1. Two versions of DIPETT compared.
"... In PAGE 10: ... In particular we looked at the percentage of full parses (as opposed to partial or fragmentary parses). Table1 presents the results. ENGLISH TECHNICAL TEXT % full parses, VERSION 3 % full parses, VERSION 4 QUIRK (578 grammatical sentences; rich in syntactic structure; phenomena interaction not controlled) 79.... In PAGE 11: ... In general, DIPETT manages to parse at least 95% of any English text, just with varying proportion of full parses. For the last five texts from Table1 , partial parsing performed in addition to the full parses listed, gives the results shown in Table 2. That is, DIPETT parsed between 91% and 99% of these texts.... ..."
Table 5: The equivalent entries between Basque and English for the verb esan.01. Note that we are showing the brief version of the entries for Basque. Figure 3 shows a full entry.
2006
Cited by 1
Table 1. Shown here for Plains, EMBOSS, and LAGAN are the r, t, and 0 values obtained from aligning genomic DNA sequences of lengths between 0:5 Kb and 12 Kb within human, mouse, dog, and fugu, where the pairs are biologically related and mainly noncoding DNA with expected large gaps and low homology regions. The conversion from to 0 was carried out for convenience in comparing lab results, where higher 0 indicates results that are less coincidental. Also, note the loss of precision involved in reporting 0 values. Hence, if for a paricular alignment, Plains and LAGAN receive 0 values that di er by less than 1 102, then their 0 values would \appear quot; equal in this table. Further information regarding the sequences used can be found in the full version of this paper at site http://bioinformatics.nyu.edu/~gill/index.shtml.
2006
"... In PAGE 6: ... The right graph in gure 2 shows the distribution of r and t values observed from randomly generated sequences after adjusting for mean and variance. From it, our approximation of T and R for P(x = T; y = R) gives us for that5: = P(x T;y R) = Z 1 T Z R 0 ece atx2+btx+cte ary2+bry+crdydx = = ec+ct+cr+ b2 t 4at + b2 r 4ar 4patar 1 Erf bt + 2atT 2pat Erf br + 2arR 2par Erf br 2par Table1 shows a comparison of alignments for biologically related sequences in terms of unadjusted r and t values, and 0 values, where = 0:5, and 0 = ln ( ). Note that Plains does not always yield the results of least coincidental probability in this table.... ..."
Cited by 3
Table 4: E cient weak blind message recovery variants Security considerations The security analysis can be adapted from the Meta-Message recovery scheme [HMP294] and will be given in the full version of the paper. E ciency considerations Below the variants with e cient signature veri cation in table 4 those are most e cient, in which the parameter ~ s occurs only as argument in the coe cient A. This is the case in variants MW I.3 and MW III.4 8. Applications Applications for message hidden signatures are those, where the notary has no interest in the message but the owner has in spite of this a special interest in the anonymity of the message. This might be the case in a testament application in which the notary signs the last will of Alice without knowing the content of it during Alice apos;s lifetime. Later the
Table 8: Dynamic Full Enumeration encoding performance
2004
"... In PAGE 9: ...able 7: Training Set analysis.............................................................................................................39 Table8 : Dynamic Full Enumeration encoding performance.... In PAGE 63: ... However, it is clear that 2x2 version encoding takes noticeably less time than 3x3 version. Table8 gives a better insight on the execution time difference: 2x2 version takes less than 10% of time that it takes for the 3x3 version to encode a data page. Table 8: Dynamic Full Enumeration encoding performance ... ..."
Table 1: RCS statistics for the FreeBSD CVS data sets. There are nearly 300,000 versions in the FreeBSD- full data set, which occupy 3.4 gigabytes of uncompressed space. The FreeBSD-min10 data set accounts for 11 percent of the files, 46 percent of the versions, and 68 percent of the volume of uncompressed data. The mean (DC) and standard deviation of the mean (AR DC ) are listed for a number of per-file statistics.
"... In PAGE 17: ...2, USENIX Edition [48], including its source tree, ports collection, documentation, and CVSROOT directory. Table1 provides statistics describing the RCS files in the FreeBSD CVS data... ..."
Table 2. Communication and Meta-data Sizes for opt Versions
"... In PAGE 10: ... We next focus on the overhead of meta-data that needs to be sent with optimized messages that are comprised of dis- joint rectangular sections. Table2 presents the total com- munication volume (in Mega Bytes), i.e.... In PAGE 10: ...ages. The size of meta-data is less than 0.03% of the to- tal communication volume in all cases. The data presented in Table2 clearly establishes that the overhead of sending meta-data with optimized messages is negligible. Table 3 shows the total communication volume for the full versions.... In PAGE 10: ... Table 3 shows the total communication volume for the full versions. These numbers can be compared against the total communication volume shown for opt versions in Table2 to see the reduction in communication volume achieved by our runtime technique. For mg-vscope, with medium query size, the reduction in communication vol- ume is 9%, 25%, and 52% on 2, 4, and 8 processors, re- spectively.... ..."
Results 11 - 20
of
22,422
