Chapter 2 generation algorithm based on division-test-extended real-value Detector

Chapter 2 generation algorithm based on division-test-extended real-value Detector

This chapter aims to improve the performance of the PT-RNSA algorithm, introduce the extension strategy on the basis of division-test, put forward the algorithm based on division-test-extended real-value detector generation (PTS-RNSA: real-valued detector generation algorithm based on the partition-test-spread proess), and compare it with the PT-RNSA algorithm and V-detector algorithm through experiments, and achieve good results.

3.1 defects of PT-RNSA Algorithm and Its Improvement Method

The PT-RNSA algorithm has the advantages of low cost of generating mature detector, no crossover of mature detector, non-I-space uniform coverage and so on. However, this algorithm also has its obvious shortcomings, mainly because, to achieve a certain detection rate, it usually requires a large number of mature detectors.

This section presents an improved idea of PT-RNSA algorithm, while maintaining the original advantages of PT-RNSA algorithm, by introducing the extension strategy, based on the Division-test-extended real-value detector generation algorithm (PTS-RNSA), without reducing the detector set coverage rate, the number of mature detectors is reduced, improves algorithm performance.

Figure 3.1 shows the core idea of PTS-RNSA.

 

1.Initialization: Regard the entire representation space as a super cube and use it as the candidate Super cube (Candidate detector ). 2.Division: Evenly divide a super cube into smaller super cubes. 3.Test: Determines whether these candidate Super cubes meet the Self set. If all candidate Super cubes are intersection with the Self set, go to step 1 to re-divide each candidate Super cube; otherwise, go to step 1. 4.Extension: For candidate Super cubes that do not interwork with the self-set, expand their coverage as much as possible. Extended Then, the coverage of each candidate Super cube is updated. For each updated candidate detector, if it does not intersect with the self-set, it is regarded as a mature detector. Otherwise, go to step 1 and continue to divide the candidate detector that interacts with the self-set.

Fig 3.1PTS-RNSA core idea

 

According to the description in Figure 3.1, the PTS-RNSA is mainly divided into three processes: division process, testing process and expansion process. Different from the PT-RNSA, The PTS-RNSA adds the extension process on the basis of the retention division process and the test process. The expansion process is mainly used to increase the coverage of mature detectors and reduce the number of mature detectors. It should be noted that after the expansion process is completed, some adjacent candidate detectors may be merged. Moreover, after the expansion process is complete, the coverage of candidate detectors that intersect with the Self set may be reduced.

Figure 3.2 shows an example of the Division-test-expansion process of PTS-RNSA in two-dimensional space, and intuitively shows the division, test and expansion idea of PTS-RNSA. The black area is the self, the blank area is not the self, the blank rectangle is the candidate detector, and the light color rectangle is the mature detector.

 

 

Figure 3.2 (a) is the initial candidate detector that interacts with the Self set. Therefore, according to the PTS-RNSA algorithm, it is evenly divided into four small rectangles, 3.2 (B ). However, these four small rectangles all intersect with the Self set, so each small rectangle is further divided. Figure 3.2 (c)-Figure 3.2 (n) describes the Division-test-expansion process in sequence.

Figure 3.2 (c)-Figure 3.2 (e) shows the small rectangle division-test-expansion process in the upper left corner. Figure 3.2 (c) is the division process, Figure 3.2 (d) is the testing process, and Figure 3.3 (e) is the expansion process.

In Figure 3.2 (c), the rectangle in the upper left corner is evenly divided into four small rectangles. In Figure 3.2 (d, the four small rectangles are tested to determine whether they are intersecting with the Self set. In Figure 3.2 (e), the coverage of the rectangles that do not overlap with the Self set is expanded as much as possible, finally, two mature detectors are generated. The rectangles that intersect with the Self set are left to be further divided after their coverage is reduced.

Figure 3.2 (f)-Figure 3.2 (H), Figure 3.2 (I)-Figure 3.2 (K), Figure 3.2 (l)-Figure 3.2 (N) describes a similar division-test-expansion process.

 

3.2 scaling policy in two-dimensional space

After a candidate detector is evenly divided into four sub-candidate detectors, it can be divided into the following four types based on the number of intersection between the sub-candidate detector and the Self-set:

A. Only one of the four sub-candidate detectors divided into is at the intersection of the Self set.

B. Two of the four sub-candidate detectors are divided into self-sets.

C. Three of the four sub-candidate detectors divided into are intersection with the self-set.

D. Four of the four sub-candidate detectors divided into are intersection with the self-set.

In Class D, the four sub-candidate detectors divided by the candidate detector all intersection with the self-set, according to The PTS-RNSA, each sub-candidate detector is divided again. In the other three categories, each category contains several small cases. In each case, an extension policy is used to make an appropriate extension.

The four subcandidate detectors are divided by candidate detectors. in clockwise direction, the four subcandidates are marked as W1, W2, W3, and w4.

The rest of this section will detail the extended solutions used for each category.

 

...................

...................

Because there are too many images, the specific analysis section is omitted here:

 

3.3 Implementation of PTS-RNSA in Two Dimensional Space

Figure 3.10 describes the pseudo-code implemented by the PTS-RNSA Algorithm in two-dimensional space. The main difference between PTS-RNSA and PT-RNSA is that the PT-RNSA performs a one-to-one test on the sub-candidate detector divided by the candidate detector. If it is not in intersection with the self-set, it is immediately seen as a mature detector, if intersection, the Division continues; while the PTS-RNSA does not immediately use these subcandidate detectors as mature detectors or divide them after testing them, instead, a scaling strategy is adopted based on the intersection of these candidate detectors and the Self-set. After the extension, a child candidate detector that does not intersection the self-set is used as a mature detector, the intersection operator continues the Division-test-expansion process.

 

 

D: Indicates a candidate detector,D= <C,R> C: The center of the detector,C= <CX,CY> R: The distance from the detector's center point to the boundary of each dimension,R= <RX,RY> R0: the predefined distance from the smallest detector center point to the boundary of each dimension D: Mature detector set Q: queue Data Structure SI (1 ≤I≤ 4): indicatesIWhether the sub-candidate detector interacts with the self-set. False indicates not intersection, and true indicates intersection. Num: Indicates that there are several sub-candidate detectors that intersect with the self-set. VI (1 ≤I≤ 4): indicatesIThe sequence of Self-individual that the sub-candidate detector intersection.

1. PTS-RNSA () 2 .{ 3.Initialization:D= <C,R> Set the initial values of candidate detectors to overwrite the entire problem representation space; The mature detector set D is left blank; Set queue Q to an empty queue; 4. If (Candidate DetectorDDo not interwork with the self-set) then 5 .{ 6. d Every {D}; 7. return; 8 .} 9. enqueue (D); // Set the candidate DetectorDPut in queue Q 10. While (Q! = NULL) 11 .{ 12.D= Dequeue (Q); // retrieve the first element from queue Q and remove it from the queue 13. PTS-detector (D); 14 .} 15 .}

16. PTS-detector (D){ 17. If (R<R0) return; 18. Associate the candidate DetectorDEvenly divided into four sub-candidate DetectorsDI (1 ≤I≤ 4 ); 19.S1 =S2 =S3 =S4 = false; 20.Num= 0; 21. For eachDI (1 ≤ I ≤ 4) 22. If (DI and self-set intersection ){ 23.SI = true; 24.Num++; 25. InVIn I, the sequence of the Self-individual that communicates with the I sub-candidate detector is stored; 26 .} 27. Switch (Num){ 28. Case 1: accordingS1,S2,S3,SWhen the value is 4, the specific extension scheme of Class A is determined. 29. Case 2: accordingS1,S2,S3,SWhen the value is 4, decide which type B extension scheme to execute 30. Case 3: accordingS1,S2,S3,SWhen the value is 4, decide the specific extension scheme of class C. 31 .} 32. For eachDI (1 ≤I≤ 4) 33. If (DI! = PHI ){ 34. If (DI is not associated with the self-set) 35. d Faster {DI }; 36. Else 37. If (DI is not completely covered by any individual) 38. enqueue (DI ); 39 .} 40 .}

Fig 3.10Pseudo-code implementation of PTS-RNSA Algorithm in two-dimensional space

 

Figure 3.10 describes the pseudocode implemented by the PTS-RNSA in a two-dimensional space.

 

 

................

................

Too much lab data is omitted here ..