### Table 1 Interference model: K-distributed clutter + noise

"... In PAGE 2: ... Sea clutter is generated using the K-distributed sea clutter model [1]. The sea clutter is combined with receiver noise, weather, landmass targets and jammers to obtain a composite interference as shown in Figure 1 and described in Table1 . The noise and clutter from each source are uncorrelated, so they are combined using root sum of squares.... In PAGE 3: ... Radar interference is modeled as the sum of Gaussian Noise and K-distributed Clutter. This interference model is defined in Table1 . A formula for the shape parameter (empirically derived in Reference [1].... ..."

### Table 2 Shape parameter for K-distributed clutter

### Table 2: Summary of the synthetic graphs Graph Parameter n k distribution

"... In PAGE 5: ... We use synthetic data to simulate graphs whose edge weights are under normal and poisson distributions. The dis- tribution parameters to generate the graphs are listed in the second column of Table2 as matrices. In a parameter matrix P, Pij denotes the distribution parameter that generates the edge weights between the nodes in the ith partition and the nodes in the jth partition.... ..."

### Table 4: Parameters used for the transmission functions due to water vapor line absorption. k1 is the rst ux absorption coe cient used with the k-distribution method. The remaining coe cients are de ned using kn = kn?1, where is the band dependent constant given in the table. g is the Planck-weighted k-distribution function de ned in (35). The use of the k-distribution method in Band 3 is described in Section 4.5. Units of k are g?1cm2; other values are non-dimensional.

"... In PAGE 23: ...Table4 . As shown in Table 1, in Bands 1, 2, and 7 water vapor line absorption can be computed using either this method or the transmittance tables described in the next subsection; in all other bands we use only the k-distribution method.... ..."

### Table 3: Scaling parameters used with the k-distribution method. Bands are de ned in Table 1. The reference temperature and pressure, Tr and pr, are those used in (24), m is the power for the pressure scaling in the same equation, and and are the temperature scaling coe cients that appear in (25).

"... In PAGE 20: ... (1993). Values of pr, Tr, m, , and used to scale the absorption coe cients for water vapor and CO2 in each of the bands are given in Table3 . Note that Band 3 is divided into three sub-bands; the reasons for this are explained in Section 4.... In PAGE 28: ... average: WV = 3 Xi=1 C i L i Bi(To) B(To) ; (51) where Bi(To) is the Planck ux integrated over sub-band i, and B(To) = P3 i=1 Bi(To) is the total Planck ux for Band 3. Substituting for the line transmittance from (34) and for the continuum transmittance from (46) we have: WV = 3 Xi=1 e?kc i ^ q= N X n=1 e?kn~ q= ( g)n;i Bi(To) B(To) = 3 Xi=1 e?kc i ^ q= N X n=1 e?kn~ q= (d g)n;i: (52) Since we use the same scaling parameters ( Table3 and (43)) for the three sub-bands, the... In PAGE 29: ... When the k-distribution method is used, we separate Sub-band 3b (the center region) from sub-bands 3a and 3c (the wings). The optical properties of Sub-bands 3a and 3c are very similar, and so we have combined them by using the same scaling parameters ( , , pr and m in Table3 ) and the same set of absorption coe cients (given by k1 and in Table 7). The band-averaged CO2 Table 7: Parameters for computing transmittances in the three sub-bands of Band 3.... ..."

### Table 1: Spectral bands, absorbers, and parameterization methods.

"... In PAGE 14: ...1 The 8 Bands For computing IR uxes due to water vapor, carbon dioxide, and ozone, we divide the spectrum into eight bands. Table1 shows the spectral ranges of these bands, together with the absorbers involved in each band and the parameterization methods used to compute the transmittance in each band. The transmittance parameterizations are discussed in Section 4.... In PAGE 19: ... The transmittances due to water vapor continuum absorption in Bands 3 through 6 are computed using a one-parameter scaling approach. The applications of these parameterizations to the di erent spectral bands and absorbers are summarized in Table1 . The code allows \HIGH quot; and \LOW quot; options to be speci ed depending on the desired accuracy in the middle atmosphere.... In PAGE 23: ...Table1 , in Bands 1, 2, and 7 water vapor line absorption can be computed using either this method or the transmittance tables described in the next subsection; in all other bands we use only the k-distribution method. In Band 3, the 15 m band, we use a somewhat more complicated procedure described in Section 4.... In PAGE 24: ... 4.3 One-Parameter Scaling For Water Vapor Continuum Absorption As shown in Table1 , water vapor continuum absorption is included in Bands 3 through 6. The water vapor continuum absorption coe cient, kc , depends on the water vapor partial... In PAGE 27: ... T = 1 2 + R B 0 1( ) 0 2( )d R B d : (49) If the overall shapes of the absorption curves due to both absorbers are uncorrelated with each other and with B , the second term on the right-hand side of (49) can be neglected and the total transmittance becomes T = 1 2: (50) Overlapping of absorption in individual bands are shown in Table1 . As shown in Chou et al.... In PAGE 27: ... Second, the CO2 absorption coe cients di er by several orders of magnitude between the band center and the wings (see the top panel of Figure 3). Rather than trying to parameterize the correlation e ect or the variations in CO2 absorption, we simply divide the band into three sub-bands (see Figure 3 and Table1 ) and then combine the parameterized transmittances of the sub-bands into a single band transmittance. This transmittance is then used in the usual way to solve the transfer equations for the entire band.... In PAGE 49: ...Table1 0: Same as Table 9, except for the sub-arctic winter atmosphere. Spectral Band (cm?1) F# sfc F quot; top 0 { 340line-by-line 40.... ..."

### Table 4. A comparison of two approaches to extending the Bayesian classifier.

"... In PAGE 22: ... This causes the Cartesian product of two discretized attributes to have 25 values, instead of 100, and leads to substantially more reliable probability estimates, given that the training set sizes are in the hundreds. The domains and training set sizes appear in the first two columns of Table4 . The remaining columns display the accuracy of the Bayesian classifier and extensions, averaged over 24 paired trials, and found by using an independent test set consisting of all examples not in the training set.... In PAGE 22: ... The remaining columns display the accuracy of the Bayesian classifier and extensions, averaged over 24 paired trials, and found by using an independent test set consisting of all examples not in the training set. In Table4 , Accuracy Once shows results for the backward stepwise joining algorithm of Pazzani (1996), forming at most one Cartesian product as determined by the highest accuracy using leave-one-out cross validation on the training set; Entropy Once is the same... ..."

### Table 2: Results on Senseval-2 English lexical sam- ple using different Bayesian network approaches.

"... In PAGE 5: ... As the dataset is used extensively for this purpose, only the Senseval-3 lex- ical sample task is used for evaluation. Selecting Bayesian Network The best achievable result, using the three different Bayesian network approaches, when validating on Senseval-2 test data is shown in Table2 . The parameters that are used... ..."

### Table 2: Results on Senseval-2 English lexical sam- ple using different Bayesian network approaches.

"... In PAGE 5: ... As the dataset is used extensively for this purpose, only the Senseval-3 lex- ical sample task is used for evaluation. Selecting Bayesian Network The best achievable result, using the three different Bayesian network approaches, when validating on Senseval-2 test data is shown in Table2 . The parameters that are used... ..."