### Table 4: k-NN Query.

2000

"... In PAGE 8: ....2.3 k Nearest Neighbor Queries A k-NN query retrieves a set of objects such that for any two objects , . The algorithm for k-NN queries is shown in Table4 . Like the ba- sic k-NN algorithm, the algorithm uses a priority queue to navigate the nodes/objects in the database in increasing order of their distances from .... ..."

Cited by 91

### Table 4: k-NN Query.

2000

"... In PAGE 14: ....2.3 k Nearest Neighbor Queries A k-NN query Q = hQ; k; Di retrieves a set R of k objects such that for any two objects O 2 R; O0 62 R, D(Q; O) D(Q; O0). The algorithm for k-NN queries is shown in Table4 . Like the basic k-NN algorithm [26, 39], the algorithm uses a priority queue queue to navigate the nodes/objects in the database in increasing order of their distances from Q.... ..."

Cited by 91

### Table 4: k-NN Query.

2000

"... In PAGE 8: ....2.3 k Nearest Neighbor Queries A k-NN query Q = hQ; k; Di retrieves a set R of k objects such that for any two objects O 2 R; O0 62 R, D(Q; O) D(Q; O0). The algorithmfor k-NN queries is shown in Table4 . Like the ba- sic k-NN algorithm, the algorithmuses a priorityqueue queue to navigate the nodes/objects in the database in increasing order of their distances from Q.... ..."

Cited by 91

### Table 4: k-NN Query.

"... In PAGE 14: ....2.3 k Nearest Neighbor Queries A k-NN query Q = hQ; k; Di retrieves a set R of k objects such that for any two objects O 2R;O 0 62 R, D#28Q; O#29 #14D#28Q; O 0 #29. The algorithm for k-NN queries is shown in Table4 . Like the basic k-NN algorithm [26, 39], the algorithm uses a priority queue queue to navigate the nodes/objects in the database in increasing order of their distances from Q.... ..."

### Table 2 summarizes the results for a KNN query with K = 10. The times presented in the table are average figures obtained from performing 100 KNN queries.

2004

"... In PAGE 24: ... Table2 : Search times for descriptors generated from topology graphs. Table 2 shows that our indexing scheme outperforms current approaches for many data distributions.... In PAGE 24: ...Table 2: Search times for descriptors generated from topology graphs. Table2 shows that our indexing scheme outperforms current approaches for many data distributions. Our indexing structure seems to scale better both with growing dimen- sionality and data set size, while exhibiting low insertion and search times, making it a good choice for interactive applications where timely feedback is required.... In PAGE 25: ... For each of the five queries, we determined the positions for the 10 similar shapes in the ordered response set. Using results from our method and the values presented in Table2 from [46] we derived the precision-recall plot shown in Figure 14. Looking at Figure 14 we can see that our technique outperforms all the other methods, yielding good precision figures for recall values up to 50%.... ..."

Cited by 3

### Table 2 summarizes the results for a KNN query with K = 10. The times presented in the table are average figures obtained from performing 100 KNN queries.

2004

"... In PAGE 24: ... Table2 : Search times for descriptors generated from topology graphs. Table 2 shows that our indexing scheme outperforms current approaches for many data distributions.... In PAGE 24: ...Table 2: Search times for descriptors generated from topology graphs. Table2 shows that our indexing scheme outperforms current approaches for many data distributions. Our indexing structure seems to scale better both with growing dimen- sionality and data set size, while exhibiting low insertion and search times, making it a good choice for interactive applications where timely feedback is required.... In PAGE 25: ... For each of the five queries, we determined the positions for the 10 similar shapes in the ordered response set. Using results from our method and the values presented in Table2 from [46] we derived the precision-recall plot shown in Figure 14. Looking at Figure 14 we can see that our technique outperforms all the other methods, yielding good precision figures for recall values up to 50%.... ..."

Cited by 3

### TABLE II RESULTS OF A KNN SKETCH QUERY.

### TABLE II RESULTS OF A KNN SKETCH QUERY.

### Table 2. Exact K-NN query results using mrkd-tree K-NN search /

2004

"... In PAGE 5: ...or this 10.5 hour MPEG1 video dataset, only less than 0.5 hour is spent for feature extraction and building index. In Table2 , CPU cost of K-NN query is given with different K. By using the NPT package [10], merely tens of milliseconds is needed to search out the very first best matching with the built index.... ..."

Cited by 11

### Table 4: Analysis of the NB-Tree k-NN Query algorithm.

2003

"... In PAGE 14: ...in Table 3 k-NN Query From the pseudo-code presented in Figure 5 we can extract the following opera- tions: computation of the Euclidean Norm, a search in the B+-Tree, the calculation of N distances (in the worst case), the insertion of M points into a list (M lt; lt; N) and N sequential steps in the B+-Tree (in the worst case). Table4 shows the running times for the k-NN algorithm.... ..."

Cited by 2