### Table 2: Comparison of EC-DSA and EC-KCDSA Note that () is a function which converts an elliptic curve point to an integer. In case of an elliptic curve de ned over GF (p), ((x1; y1)) = x1. In case of GF (2m), since x1 can be represented as a binary string (sm?1sm?2 s1s0), ((x1; y1)) = Pm?1 i=0 si2i.

"... In PAGE 7: ... It is only needed at the time of key pair generation. For comparison, in DSA a multiplicative inverse mod q needs to be evaluated each time a signature is generated or veri ed and in GOST each time a signature is veri ed (see Table 1 and Table2 ). Evaluating an inverse mod q would take very little portion in the overall workload of signing/verifying on most general purpose computers.... ..."

### Table 1: Probabilistic Approaches

"... In PAGE 2: ...3 Word-based, Probabilistic Approaches The third category assumes at most whitespace and punctuation knowledge and attempts to infer MWUs using word combination probabilities. Table1 (see next page) shows nine commonly-used probabilistic MWU-induction approaches. In the table, f and P signify frequency and probability XX of a word X.... ..."

### Table 2. Error rates obtained by Probabilistic Neural Networks .

"... In PAGE 3: ... Its architecture is very simple and powerful, and has some advantages when compared to MLPs: rapid training speed, PNNs are asymptotically Bayes optimal if enough training samples are provided, a good performance over discontinuities in the input space, and what is much more important, it provides mathematically confidence levels for its decisions. Table2 shows the results for PNN in the estimation of thickness and defocus using the same testing procedure than in the previous paragraph. The combinations of both sets of features using PNNs gave the best results in the artificial experiments, and were selected as the definite choice to be considered in real experiments.... ..."

### Table 1. Probabilistic Cue-criterion Relationships Linking Aircraft Characteristics (cue values) to Aircraft Identities (criterion values)

"... In PAGE 2: ... Cue values (one of three possible textual values) were presented in a fixed sequential order (first IFF, then radar, altitude, and origin), as participants asked for them, because the rule verification task assumed that participants had experienced IFF values and their occurrence with hostile or friendly aircraft. Table1 gives the probabilities that were used to define the types of aircraft which occurred during the experimental scenarios, as well as to define the probabilities of finding certain types of evidence for the rule verification task. Participants performed eight, ten-minute experimental sessions, in which they had to identify 25 aircraft which... ..."

### Table 2: Overview of flexible multiplication units supporting several fields in GF(2m).

"... In PAGE 5: ... It is premised that smaller polynomials are padded with zeros to meet the full multiplication factor length. Table2 shows an overview of multipliers for different fields applying the reduction methods as they were described in the previous section. One can see that the MHWR approach is very efficient.... ..."

### Table 4 - Clock cycles to perform modular multiplication in GF(p) or GF(2m) (radix 8)

"... In PAGE 18: ... The total number of cycles to perform a modular multiplication on n-bit operands will depend on the kernel configuration. Table4 shows the number of clock cycles to perform the operation for some typical n values and some chosen kernel configurations (with approximately 30,000 gates). Observe that the hardware is capable of executing in 0.... ..."

### Table X.6: GF(2m) elliptic curve cryptosystem hardware comparison

### Table 6.4: Area estimates for scalar GF(2m) multiply-accumulate units.

2005

### Table 14. Verification of SBSNNI on Access Monitor 5 with the compositional algo- rithm.

"... In PAGE 58: ... We could hope to get a better result using the composi- tional algorithm. Table14 reports the output of the compositional verification of Access Monitor 5 where the symbols [\] and [|] represent steps 1 and 2 of the algorithm, respectively. This table shows that the algorithm fails in the verification of SBSNNI over AM and then succeeds in checking the system as a whole.... ..."