### Table 1: Music metadata fields, with some example values

2002

"... In PAGE 3: ... This music metadata vector is assigned by editors to every track of a large corpus of music CDs. Sample values of these variables are shown in Table1 . Our kernel function C3B4DCBN DCBCB5 thus computes the similarity between two metadata vectors correspond- ing to two songs.... In PAGE 6: ...Table1 . The resulting kernel is still defined by (4), with a specific AWD2 that will be defined in section 3.... In PAGE 6: ... We could convert the categorical data to a vector space by allocating one dimension for every possible value of each categor- ical variable, using a 1-of-N sparse code. This would lead to a vector space of dimension 1542 (see Table1 ) and would produce a large number of kernel parameters. Hence, we have designed a new kernel that operates directly on categorical data.... In PAGE 6: ... Due to space limitations, proof of the Mercer property of this kernel is omitted. For playlist generation, the AWD2 operate on music metadata vectors DC that are defined in Table1 . These vectors have 7 fields, thus D0 runs from 1 to 7 and D2 runs from 1 to 128.... ..."

Cited by 11

### Table 1: Music metadata fields, with some example values

"... In PAGE 3: ... This music metadata vector is assigned by editors to every track of a large corpus of music CDs. Sample values of these variables are shown in Table1 . Our kernel function C3B4DCBN DCBCB5 thus computes the similarity between two metadata vectors correspond- ing to two songs.... In PAGE 6: ...Table1 . The resulting kernel is still defined by (4), with a specific AWD2 that will be defined in section 3.... In PAGE 6: ... We could convert the categorical data to a vector space by allocating one dimension for every possible value of each categor- ical variable, using a 1-of-N sparse code. This would lead to a vector space of dimension 1542 (see Table1 ) and would produce a large number of kernel parameters. Hence, we have designed a new kernel that operates directly on categorical data.... In PAGE 6: ... Due to space limitations, proof of the Mercer property of this kernel is omitted. For playlist generation, the AWD2 operate on music metadata vectors DC that are defined in Table1 . These vectors have 7 fields, thus D0 runs from 1 to 7 and D2 runs from 1 to 128.... ..."

### Table 1. Comparison of limiting gravity-wave properties with Mercer amp; Roberts #281992#29. The

"... In PAGE 6: ... The #0Cnite- di#0Berence prediction at 1=,=1=30 #28where the solid and dashed lines meet in #0Cgure 2#29 is within graphic accuracy of the spectral method prediction. Table1 shows that the maxi- mum values of wave height, energy, and period obtained with a 64-node mesh agree with Mercer amp; Roberts #281992#29 to three signi#0Ccant digits. A convergence study shows that our values are accurate to at least the digits shown for the maximum values.... In PAGE 9: ... These studies could not prove or disprove the 90 o conjecture as their choice of expansion parameter ka reaches a maximum before the extremal values of the crest curvature and the crest angle are reached. Using a denser node distribution at the crest #280:5 #3CS#3C0:8#29 improves our calculation beyond the steepest wave calculated by Mercer amp; Roberts #281992#29 #28 Table1 #29. In #0Cgure 2, the wave slope increases rapidly beyond one as the crest radius #281=,#29 approaches zero, while the crest acceleration A c approaches one.... ..."

### Table-2. Results for AP88 collection (for Jelinek-Mercer smoothing).

2006

Cited by 1

### Table 1: Music metadata fields, with some example values

2002

"... In PAGE 3: ... This music metadata vector is assigned by editors to every track of a large corpus of music CDs. Sample values of these variables are shown in Table1 . Our kernel function a0a2a1a4a3a6a5a7a3 a8 a10 thus computes the similarity between two metadata vectors correspond- ing to two songs.... In PAGE 6: ...Table1 . The resulting kernel is still defined by (4), with a specific a65a57a62 that will be defined in section 3.... In PAGE 6: ... We could convert the categorical data to a vector space by allocating one dimension for every possible value of each categor- ical variable, using a 1-of-N sparse code. This would lead to a vector space of dimension 1542 (see Table1 ) and would produce a large number of kernel parameters. Hence, we have designed a new kernel that operates directly on categorical data.... In PAGE 6: ... Due to space limitations, proof of the Mercer property of this kernel is omitted. For playlist generation, the a65 a62 operate on music metadata vectors a3 that are defined in Table1 . These vectors have 7 fields, thus a22 runs from 1 to 7 and a24 runs from 1 to 128.... ..."

Cited by 11

### Table: Evaluation on manually annotated CELEX and a grapheme-to- phoneme conversion task using the Add-WordTree decision tree (Lucassen and Mercer [1984]).

2007

Cited by 4

### Table 1: New system calls for SRT tasks. API Description

2003

"... In PAGE 6: ... Adding new system calls. Weaddthreenewsys- tem calls ( Table1 ) to support soft real-time requirements of multimedia tasks. Atask uses start_srt to declare itself as a SRT task and to require statistical performance guaran- tees from the OS.... In PAGE 12: ...5% 20.6% wrsUni Table1 0: Energy and deadline miss ratio for con- current run. 4.... ..."

Cited by 63

### Table 15 Best SVMR Predictions for the Bioconcentration Factors of 238 Diverse Organic Compoundsa

"... In PAGE 80: ...Table15 contains the best SVM regression results for each kernel. The cross-validation results show that the correlation coefficient decreases in the fol- lowing order of kernels: linear gt; degree 2 polynomial gt; neural gt; RBF gt; anova.... ..."

### Table 13 Best SVMR Predictions for the Benzodiazepine Receptor Ligands QSARa

"... In PAGE 77: ...Table13 , we present the best regression predictions for each kernel. Despite the large number of SVMR experiments we carried out for this QSAR (34 total), the cross-validation statistics of the SVM models are well below those obtained with MLR.... ..."

### Table 14 Best SVMR Predictions for the Toxicity QSAR of Aromatic Compounds to Chlorella vulgarisa

"... In PAGE 78: ... The neural network does not improve the prediction of log 1/EC50 compared with the MLR model. The best SVM regression results for each kernel are given in Table14 . By comparing the results from MLR, ANN, and SVMR, we find that no clear Table 14 Best SVMR Predictions for the Toxicity QSAR of Aromatic Compounds to Chlorella vulgarisa... ..."