### Table 1: Bayesian inference of singular values, lr, given different rank estimates: Bayesian selection (r = 3), and ARD Property (r = 9). Bayesian selection ARD Property

"... In PAGE 15: ... = 10. True (i.e. the simulated) singular values are given in Table1 together with approximating moments of their posterior distribution (39). Simulated values are clearly within the uncertainty bounds.... In PAGE 15: ... Moments of the transformed distri- butions are displayed in Table 2 and Table 3, together with the projection of the original simulated values in each case. Again, we condition on the two cases of r used in Table1 . Projected true values of Ar and Xr are, in both cases, within the uncertainty bounds of the posteriors (A.... ..."

### Table 2 tabulates the experimental results obtained from both corpora using Bayesian Networks, Support Vector Machines and our hybrid algorithm of using Bayesian inference knowledge and inserting it into the SVM hyperparameters, noted as B-SVM. For the task of identifying a valid SF using additional unlabeled examples, we have conducted the experiments using approximately 6.500 unlabeled instances from each corpus. Additionally, in order to obtain a more inclusive view of the task, we provide results using statistical machine learning algorithms such as relative frequency

2002

"... In PAGE 7: ...-2+3] 85.7 89 93.4 80.3 72.6 78.7 Table2 : Experimental results obtained from the two corpora By observing the obtained results, we could claim that both BBN and SVM perform significantly better than the other machine learning algorithms by a factor that varies from 5 to almost 30%, a fact that supports the argue that BBN and SVM are well suited for the task of verb subcategorization identification (Maragoudakis et al, 2001). Furthermore, by incorporating bayesian knowledge into the SVM classifier and using a set of unlabeled examples, we achieve a 3-6% improvement.... ..."

Cited by 1

### Table 3: Inference of Xr via OVPCA(displayed in transformed variable yX;r). Bayesian selection ARD Property

"... In PAGE 15: ...e exploited in the rapid evaluation of (77) as discussed in the second bullet point of Section 6.1. VB approximating posterior distributions for orthogonal parameters Ar (37) and Xr (38) are presented in transformed variables, yA and yX (78), respectively. Moments of the transformed distri- butions are displayed in Table 2 and Table3 , together with the projection of the original simulated values in each case. Again, we condition on the two cases of r used in Table 1.... ..."

### Table 2: Approximate AI step for Bayesian networks

"... In PAGE 5: ...Table 2: Approximate AI step for Bayesian networks Table2 describes the approximate AI step in greater de- tail. The maximization in step 7 of the algorithm is a lo- cal computation: the value CE(Pct;j 1; P t) is given at this point.... ..."

### Table 2: Parallel Bayesian Inference in a Tree Network.

1998

"... In PAGE 5: ... variable, etc. The full psuedocode is given in Table2 . For any constant number of evidence variables, its runningtime is O(log n) with n processors.... ..."

Cited by 12

### TABLE III The inference results of fatigue Bayesian network model

2004

Cited by 5

### Table 1. Steps in Mixture Simulation and Conventional Bayesian Inference

"... In PAGE 3: ... Generate an observation, x2, from the posterior. Table1 lists the steps in the mixture simulation approach alongside major steps... ..."

### TABLE I UTTERANCE LEVEL BAYESIAN ADAPTIVE INFERENCE PERFORMANCE

### TABLE III INCREMENTAL BAYESIAN ADAPTIVE INFERENCE PERFORMANCE ON THE COMPLETE DATA SET

### Table 2. Approximate estimates of the regression variance for homogenous Bayesian model

"... In PAGE 15: ...ate posterior distributions of the average squared residuals (i.e. approximate estimates of the regression variances) obtained from the actual data. The means of these distributions for the age group 24 are shown in Table2 . To try to capture the largest differences between the age groups, we decided on a model with 10 different variance parameters.... ..."