### Table 2 The adjustable parameters used for the reconstruction of the exocytotic dynamics Reaction Variable Search range

2006

"... In PAGE 4: ....4. Genetic algorithm In cases of high-dimensional optimization problems with possible non- smooth fitness function and multiple local minima, the optimization of the fitness function can be performed by the application of the GA. In the present study, each generation consisted of 100 genotypes , each of them having a random set of adjustable parameters that were selected within the permitted range (see Table2 ). The program used these values to reconstruct several experimental signals, and to calculate the fitness.... In PAGE 6: ...constant of the activation of Munc13 by calcium (step 0). It converged to a very narrow range between 300 and 600 nM ( Table2 ). On the other hand, the rate constant for the activation of Munc13 by calcium (k0) did not converge.... In PAGE 7: ...hich we used only part of the constraints (Fig. 1 and Fig. 3). The range of the variation for each adjustable parameter is summed up in Table2 , column 4. The best 10 solutions were selected and the mean values plus standard deviations for each adjustable parameter were calculated (see Table 2, columns 5 and 6).... In PAGE 7: ...hich we used only part of the constraints (Fig. 1 and Fig. 3). The range of the variation for each adjustable parameter is summed up in Table 2, column 4. The best 10 solutions were selected and the mean values plus standard deviations for each adjustable parameter were calculated (see Table2 , columns 5 and 6). The robustness [38] of the model demands that a random combination of values falling within the limits of the range of variance (column 4 in Table 2) will generate curves that retain the general features of the dynamics.... In PAGE 7: ... The best 10 solutions were selected and the mean values plus standard deviations for each adjustable parameter were calculated (see Table 2, columns 5 and 6). The robustness [38] of the model demands that a random combination of values falling within the limits of the range of variance (column 4 in Table2 ) will generate curves that retain the general features of the dynamics. This was tested by generating 200 combinations of parameters that were randomly selected from the range given in column 4 in Table 2,andthe corresponding shape of the reconstructed signals was checked.... ..."

### Table 6 Multiplierless 32d QMF filter banks found by DLM with static and adaptive weightsa

1998

"... In PAGE 23: ...ll the coe cients to be 0.9474. After multiplying each coe cient by this scaling factor and restricting each coe cient to a PO2 form with a maximum of 6 ONE bits, we apply DLM with both static and dynamic weights to find the best PO2 designs. Table6 compares the objectives of the designs found and the corresponding convergence times of DLM. Using the adaptive DLM, all the searches con- verge, and most designs have better reconstruction errors.... ..."

### Table 2: Squared correlation coefficients for network reconstructions.

"... In PAGE 25: ... These probabilities were then interpreted as predictions of relative importance for each event in the distributed representation. The probability measures of relative importance at the whole-tune level were correlated with the musical improvisation data, as summarized in Table2 . The correlations were large for the known (p lt;.... In PAGE 25: ...10). Next, we compared the network measures of relative importance with the quantifications of theoretical predictions, as shown in Table2 . The correlations with time-span reduction predictions were significant for the known and variant melodies and for the measure-level reconstruction of the novel melody (p lt;.... ..."

### Table 2: Squared correlation coefficients for network reconstructions.

"... In PAGE 25: ... These probabilities were then interpreted as predictions of relative importance for each event in the distributed representation. The probability measures of relative importance at the whole-tune level were correlated with the musical improvisation data, as summarized in Table2 . The correlations were large for the known (p lt;.... In PAGE 25: ...10). Next, we compared the network measures of relative importance with the quantifications of theoretical predictions, as shown in Table2 . The correlations with time-span reduction predictions were significant for the known and variant melodies and for the measure-level reconstruction of the novel melody (p lt;.... ..."

### Table 1. Network reconstruction performance in terms of AUC

"... In PAGE 5: ... Results are summarized in Tables 1 and 2. We see in Table1 that in terms of AUC, the local model performs on average as well as all of the non-direct methods. In particular, Table 1.... ..."

### Table 1: Performance on reconstruction of the yeast metabolic networks.

2007

"... In PAGE 4: ...ac.jp/~yoshi/ismb04 Table1 shows the performance of each pairwise kernel, as well as that of the baseline direct approach, for the differ- ent data sets. The MLPK is never worse than the TPPK ker- nel, and both methods are always much better than the baseline direct method for edge inference.... ..."

### Table 7: Dynamic network environments

1996

"... In PAGE 24: ... The network starts out with one permanently congested node and another temporar- ily congested node. Table7 summarizes the network conditions we experiment with. The local arrival rate of the permanently congested node is 0:8 and that of the temporarily congested node is 0:8 during the interval (100,400) and is equal to 0:2 at other times.... In PAGE 24: ...Table 7: Dynamic network environments Table 8 shows the behavior of the dynamic protocol in the dynamic network settings shown in Table7 . The table shows the time at which a split or a merge occurs.... ..."

Cited by 34

### Table 8: Behavior of the dynamic protocol in dynamic network environments

1996

"... In PAGE 24: ... cong. node #1 #1 16 1 #2 #1 16 6 #3 #2 6 1 #4 #2 6 3 #5 #3 1 5 #6 #3 1 10 Table 7: Dynamic network environments Table8 shows the behavior of the dynamic protocol in the dynamic network settings shown in Table 7. The table shows the time at which a split or a merge occurs.... ..."

Cited by 34

### Table 2. Robustness of Neurogenic Network Models to Variation in Initial Conditions, Prepattern

2002

"... In PAGE 3: ... each of the focal cells. For all three networks, we found solutions that were robust to this noise ( Table2 ). Simi- larly, both the augmented and reduced networks can The reduced, standard, and augmented versions of the select a single winner with very little initial bias in the network have 53, 63, and 69 such parameters, respec- prepattern by amplifying small amounts of noise (Table tively.... ..."

Cited by 7

### Table 1. Robustness of Neurogenic Network Models to Parameter Variation

2002

"... In PAGE 3: ... Any set of randomly chosen parameter values that allows the model to pass some functional test we call a solution . Allthreemodelsexhibitedlateralinhibitioninthe2-cell test, but the standard model had a much lower fre- quency of solutions than the augmented model, while the reduced model had the highest solution frequency of thethree ( Table1 ).By this naive measure,the reduced model seems the most robust of the three.... In PAGE 3: ... However if p is the probability of picking a good value for each parameter independently, we expect that the solution frequency for models with equivalent robustness per parameter would scale as p^n, where n is the number of parameters. p is highest for the augmented model ( Table1 ). That measure makes the augmented network most robust.... In PAGE 3: ... That measure makes the augmented network most robust. All three networks could also mediate lat- eral inhibition between a central neural cell and six sur- rounding cells (7-cell pattern, Figure 2), with similar solu- tion frequencies as for the 2-cell pattern ( Table1 ). As with the segment polarity network [9], cooperativity (Hill coefficients H11022 1) in transcriptional activation and repres- sion is essential for the network to function (Table 1).... In PAGE 3: ... We tion of N activation (red lines in Figure 1). Both are sup- run the model for 300 min and measure whether the initially higher portedbyexperiments,andbothcontributetoimproving cell(s) achieve(s) a high concentration of AC (greater than 20% of the solution frequency ( Table1 ), but E(SPL) autoinhibi- the highest steady-state concentration possible) and whether the initially lower cell(s) turn off (below 2% of maximum steady state). tion contributes more than cis-inhibition of N by DL.... In PAGE 5: ... But if differences in Dl levels are conducted another random parameter search inside important to the mechanism, how can the process still these newrestricted ranges. The successrate increased succeed in most cases despite constitutive Dl expres- almost 1,000-fold inside these (still very broad) ranges sion [13]? ( Table1 ). Using the same restricted range found with We added a constitutive input (with tunable parame- the 2-cell pattern, the success rate with the 7-cell test ters determining that input rate) to the dl promoter in also increased more than 100-fold.... In PAGE 6: ...33 0.25 ofoffspringsetsatpassingourtestsgenerallycorrelates with the frequencies of finding solutions through a ran- dom search of parameter space ( Table1 ). As with ran- The Shape of the Working Region dom sampling, the recombination success rate was of Parameter Space much higheramong solutions found withinthe restricted Clearly, simple but realistic representations of the core parameter ranges.... ..."

Cited by 7