Recently bit minwise hashing has been applied to largescale learning and sublinear time near neighbor search The major drawback of minwise hashing is the expensive pre processing as the method requires applying eg 200 to 500 permutations on the dat ID: 26381
Download Pdf The PPT/PDF document "One Permutation Hashing Ping Li Departme..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
0123456789101112131415 0 1 0 1 00 1 000 1 01000 1 00 11 000 1234 00 1011 1011 11 00 0000 01 IndexData Points 11 01 6 ,15, 26, 7933, 4897, 49, 2083, 14, 32, 9711, 25, 998, 159, 331 00 1011 1011 11 00 0000 01 IndexData Points 11 01 (empty) 6 , 110, 143 3, 38, 217 5, 14, 20631, 74, 153 21, 142, 329 0 2000 4000 6000 8000 10000 0 1 2 3 4x 104 # nonzerosFrequency Webspam 10-3 10-2 10-1 100 101 102 80 82 84 86 88 90 92 94 96 98 100 CAccuracy (%) b = 1 b = 2 b = 4,6,8 logit: k = 512 Webspam: Accuracy Original 1 Perm k Perm 10-3 10-2 10-1 100 101 102 80 82 84 86 88 90 92 94 96 98 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6,8 logit: k = 256 Webspam: Accuracy Original 1 Perm k Perm 10-3 10-2 10-1 100 101 102 80 82 84 86 88 90 92 94 96 98 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6,8 logit: k = 128 Webspam: Accuracy 10-3 10-2 10-1 100 101 102 80 82 84 86 88 90 92 94 96 98 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6,8 SVM: k = 256 Webspam: Accuracy Original 1 Perm k Perm 10-3 10-2 10-1 100 101 102 80 82 84 86 88 90 92 94 96 98 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6,8 SVM: k = 128 Webspam: Accuracy Original 1 Perm k Perm 10-3 10-2 10-1 100 101 102 80 82 84 86 88 90 92 94 96 98 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6 b = 8 SVM: k = 64 Webspam: Accuracy 10-3 10-2 10-1 100 101 102 80 82 84 86 88 90 92 94 96 98 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6 b = 8 logit: k = 64 Webspam: Accuracy 10-3 10-2 10-1 100 101 102 80 82 84 86 88 90 92 94 96 98 100 CAccuracy (%) b = 1 b = 2 b = 4,6,8 SVM: k = 512 Webspam: Accuracy Original 1 Perm k Perm 10-1 100 101 102 103 50 55 60 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6 b = 8 SVM: k = 32 News20: Accuracy 10-1 100 101 102 103 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6,8 logit: k = 1024 News20: Accuracy 10-1 100 101 102 103 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6 b = 8 logit: k = 256 News20: Accuracy 10-1 100 101 102 103 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6 b = 8 logit: k = 512 News20: Accuracy 10-1 100 101 102 103 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6 b = 8 SVM: k = 512 News20: Accuracy 10-1 100 101 102 103 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6 b = 8 SVM: k = 256 News20: Accuracy 10-1 100 101 102 103 50 55 60 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6 b = 8 SVM: k = 128 News20: Accuracy 10-1 100 101 102 103 50 55 60 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6 b = 8 SVM: k = 64 News20: Accuracy 10-1 100 101 102 103 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6,8 SVM: k = 2048 News20: Accuracy Original 1 Perm k Perm 10-1 100 101 102 103 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6,8 SVM: k = 1024 News20: Accuracy 10-1 100 101 102 103 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4,6,8 logit: k = 4096 News20: Accuracy Original 1 Perm k Perm 10-1 100 101 102 103 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6,8 logit: k = 2048 News20: Accuracy Original 1 Perm k Perm 10-1 100 101 102 103 50 55 60 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6 b = 8 logit: k = 32 News20: Accuracy 10-1 100 101 102 103 50 55 60 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6 b = 8 logit: k = 64 News20: Accuracy 10-1 100 101 102 103 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4,6,8 SVM: k = 4096 News20: Accuracy Original 1 Perm k Perm 10-1 100 101 102 103 50 55 60 65 70 75 80 85 90 95 100 CAccuracy (%) b = 1 b = 2 b = 4 b = 6 b = 8 logit: k = 128 News20: Accuracy