### Table 1. Separation errors of the tested robust methods.

### Table 2: Hazard rate estimation results

"... In PAGE 11: ... This implies that we can proceed with the Cox proportional hazards framework as expressed in equation (2). The maximum likelihood ratio estimates of the parameter vector b and the parameters of the baseline hazard rate can be found in Table2 . The results were computed using the STREG- procedure of Stata 6.... In PAGE 11: ... We fail to find evidence for this relation to have a quadratic component. Leaving this aside for a moment, the estimates presented in the last column of Table2 suggest that the hazard rate would decrease up to a relative interest rate of (-26.... ..."

### Table 4: Marginal Frequencies of Model Selection Based on 5050 Simulations non-robust robust methods

"... In PAGE 18: ... Note that in computation all non-robust criteria here can be regarded as special situations of the robust criteria by setting the tuning parameter c equal to 1 and the weight function w(x) = 1. The simulation results for the model selection are shown in Table4 and Figures 3 to 10. In each bar-plot in Figure 3 to Figure 10, there are 101 bars.... In PAGE 19: ...Table4 gives a measure of overall performance against an outlier of y by di erent methods. Figures 3 to 10 and Table 4 support the following arguments: (1).... In PAGE 19: ...of y by di erent methods. Figures 3 to 10 and Table4 support the following arguments: (1). The robust methods give a signi cant improvement over non-robust methods in terms of the number of times of selecting correct models.... ..."

### Table 3: Model performance for linear regressors (lms and l1s) and robust methods (rregs and rlms).

"... In PAGE 10: ...Table 3: Model performance for linear regressors (lms and l1s) and robust methods (rregs and rlms). Table3 shows each of the models that was obtained, parametrized over LF selection methods and the type of regressor. The model coe cients b0 and b1 are in- dicated as computed from the full data set.... In PAGE 10: ... It would be interesting to apply a Bayesian procedure to select the intercept from given priors. Table3 shows that in this data set the Generalization strategy is consistently better, for all models and for both the predictive error measures NMSE and NMAE. Method NMSE Comments rreg-default-G 0.... ..."

### Table 8: Robustness results Method Ratio Local Ratio Cost

2002

"... In PAGE 12: ...3 Additional Results In this section, we examine the effects of more than one agent defecting to Knapsack simultaneously, as well as the effects of varying the valuations of the secondary bidders away from the 75% level. Table8... In PAGE 13: ... In aggregate, the results are almost identical to scenarios where all agents use Knapsack. The remaining rows of Table8 show the results of varying the secondary bidder valuation from the 75% used in the preceding experiments. The values for the run with secondary bidder valuations at 75% are repeated from Table 7.... ..."

Cited by 6

### Table 4. Comparative study of robust solutions and their associated loss values, as obtained by the different robust-design methods

2004

"... In PAGE 10: ... (vi) a dual-response method: constructing a response model for the loss function, which is based on both the expected value of the response and its variance and optimizing this model explicitly (Myers and Mont- gomery, 1995). Table4 demonstrates the proposed solutions. It compares the real optimal solution (had the model in Equation (17) been known) to the solutions of the suggested approach and of the traditional methods.... In PAGE 10: ... The loss values of the other methods are scored as a percentage of the real loss and are higher than 100%. Note from Table4 , that the suggested approach yields a solution which is very close to the real optimal solu- tion. Moreover, it obtains the smallest loss among all other traditional methods (with a score of 100.... ..."

### Table 4: LM Feedback amp; Additive Effect Method ROBUST04 ROBUST05

2006

"... In PAGE 7: ... It would be in- teresting to compare their performance. In Table4 , we re- port the performance of the model-based feedback method in language modeling approaches [30]. Internal PFB (IPFB) is the pseudo feedback method.... ..."

Cited by 5

### Table 1. Comparison of conventional and robust regression methods using hypothetical data and data with one outlier. Detected outliers are italicised

"... In PAGE 1: ...90 + 1.69~ and the calibration results are given in Table1 . The residuals listed show that the LS line is attracted strongly by this single outlier and therefere fits the Table 1.... In PAGE 2: ... 2. Illustration of SM for the data in Table1 . The value of a1 versus the ranks of the slopes of all pairs of points; 4-5 indicates the slope of the line between the fourth and the fifth data point, etc.... In PAGE 3: ... The breakdown point is therefore 50%. To illustrate the method the data in Table1 are again used. Firstly, the slopes of each of the cn2 combinations of pairs of points are calculated, then the squared residuals towards the line for each measuring point are calculated, the resulting squared residuals are sorted and their medians obtained.... In PAGE 3: ... 3. Illustration of RM for the data in Table1 . (a) Ranked slope al for each point i, joined by a line to each of the other points; and (b) ranked median slopes selected from (a).... In PAGE 3: ... 4. Illustration of LMS for the data in Table1 . The ranked log of the median of squared residuals for the lines through the different ... In PAGE 6: ... Such an example is shown in Fig. 9, where the objective function (or rather its inverse for graphical purposes) is given for the data of Table1 as a function of a. and al.... ..."