Results 1 - 10 of 14,153

Table 1: Relation Set Count (Total Counts include ex- amples that yielded semantic representations for both EDUs)

in Discourse Parsing: Learning FOL Rules based on Rich Verb Semantic Representations to automatically label Rhetorical Relations
by Rajen Subba
"... In PAGE 6: ... As we add more training data in the future, we will see if rules that are more elaborate than the ones in Figure 6 are learned . 4 Evaluation of the Discourse Parser Table1 shows the sets of relations for which we managed to obtain semantic representations (i.e.... ..."

Table 9 In the original representation of the task, some of the effects of the first action were delayed by one step. Therefore, a greedy approach will yield a suboptimal reward value of 6. However, in the Walsh canonical representation, this effect is removed. A greedy approach here yields the optimal reward value of 16. As this would suggest, examination of the Walsh coefficients (shown in

in Analytical Design of Reinforcement Learning Tasks
by Robert E. Smith

Table 1 - Comparison of representations for shortest path finding

in NETWORK REPRESENTATION AND SHORTEST PATH REFLECTING THE RAMP ENTRY OR EXIT DIRECTION LIMITATION
by Kazutaka Takao, Tohru Higashi, Koji Yasuda, Yasuo Asakura
"... In PAGE 6: ...A comparison of representations for shortest path finding is shown in Table1 . As described above, the conventional representation has the possibility of yielding incorrect shortest paths.... ..."

Table 2: Avatar representation matrix

in The Effect of Avatar Connectedness on Task Performance
by John M. Linebarger, G. Drew Kessler
"... In PAGE 3: ... 4. Avatar Representations As Table2 indicates, the two crossed avatar factors, connection and correlation (implemented here as embodiment and color-coding), yielded four different avatar representations. The unconnected representations simply reflected the position and orientation information from the trackers using an appropriate icon.... ..."

Table 1: The e ects of system-internal combination by using di erent output representations. A straight-forward majority vote of the output yields better bracket accuracies and F =1 rates than any included individual classi er. The bracket accuracies in the columns O and C show what percentage of words was correctly classi ed as baseNP start, baseNP end or neither.

in Erik F. Tjong Kim Sang
by Walter Daelemans Herv'e, Erik F. Tjong, Kim Sang, Walter Daelemans, Rob Koeling, Yuval Krymolowski, Vasin Punyakanok, Dan Roth
"... In PAGE 5: ... These learning algorithms have pro- cessed ve di erent representations of the out- put (IOB1, IOB2, IOE1, IOE2 and O+C) and the results have been combined with majority voting. The test data results can be found in Table1 . In all cases, the combined results were better than that of the best included system.... ..."

Table 1: The e ects of system-internal combination by using di erent output representations. A straight-forward majority vote of the output yields better bracket accuracies and F =1 rates than any included individual classi er. The bracket accuracies in the columns O and C show what percentage of words was correctly classi ed as baseNP start, baseNP end or neither.

in Applying System Combination to Base Noun Phrase Identification
by Erik F. Tjong, Kim Sang, Walter Daelemans, Hervé Déjean, Rob Koeling, Yuval Krymolowski, Vasin Punyakanok, Dan Roth
"... In PAGE 5: ... These learning algorithms have pro- cessed ve di erent representations of the out- put (IOB1, IOB2, IOE1, IOE2 and O+C) and the results have been combined with majority voting. The test data results can be found in Table1 . In all cases, the combined results were better than that of the best included system.... ..."

Table 1: The e ects of system-internal combination by using di erent output representations. A straight-forward majority vote of the output yields better bracket accuracies and F =1 rates than any included individual classi er. The bracket accuracies in the columns O and C show what percentage of words was correctly classi ed as baseNP start, baseNP end or neither.

in Proceedings
by Th International Conference, Erik F. Tjong, Kim Sang, Walter Daelemans, Rob Koeling, Yuval Krymolowski, Vasin Punyakanok, Dan Roth
"... In PAGE 5: ... These learning algorithms have pro- cessed ve di erent representations of the out- put (IOB1, IOB2, IOE1, IOE2 and O+C) and the results have been combined with majority voting. The test data results can be found in Table1 . In all cases, the combined results were better than that of the best included system.... ..."

Table 3 lists all tense combinations that yield acceptable

in A Simplified Theory of Tense Representations and Constraints on Their Composition
by Michael R. Brent

Table 6 { Leave-One-Out Results on the Sonar Database Using the 126-Feature \Boundary- Overlapping quot; Representation Classi er With Scaled Data With Raw Data

in Case-Based Sonogram Classification
by David Aha, Patrick Harrison 1994
"... In PAGE 9: ... Similarly, feature i contains the sum of the values from frequency boundaries i, i + 1, and i + 2, where i ranges between zero and 125. Table6 summarizes the results when using this 126-feature representation. In general, this boundary-overlapping representation did not yield higher performances.... ..."
Cited by 1

Table 3. Categorization of a 400-warrior sample from generation 4 of the CCAI dataset, with class counts and accuracy for the score and combined representations

in Automatic Categorization of Human-coded and Evolved CoreWar Warriors
by Doni Pracner, Mirjana Ivanović
"... In PAGE 7: ... Imp-containing pa- pers, however, were not nearly as well optimized and rarely benefited from the presence of defensive imp structures. The following classifiers were tested: SMO, MLP, BayesNet, and IBk, with the results summarized in Table3 . It can be seen that the introduction of static features to the score-based representation does not yield consistent improvements as with the h1c dataset.... ..."
Results 1 - 10 of 14,153