400 likes | 521 Views
Kernel Methods for Implicit Surface Modeling B. Scholkopf, J. Giesen, S. Spalinger. r92922120 黃邦洪 r93922020 楊惠菁 r93922038 柯政宏 r93944009 劉弘偉. Outline. Introduction Single-Class SVMs Slab SVMs Epsilon-SVR Implement Result. Introduction. Step 1:
E N D
Kernel Methods for Implicit Surface ModelingB. Scholkopf, J. Giesen, S. Spalinger r92922120 黃邦洪 r93922020 楊惠菁 r93922038 柯政宏 r93944009 劉弘偉
Outline • Introduction • Single-Class SVMs • Slab SVMs • Epsilon-SVR • Implement • Result
Introduction • Step 1: • Given a finite sample points belong to some hypersurface in Euclidean space R3 • The sample can be very large, noisy or sparse
Introduction • Step 2: • Use SVM to find an implicit surface • f is the decision function in SVM • The zero set of f is a surface that approximates the sample points • Single-Class SVM v.s. Slab SVM
Outline • Introduction • Single-Class SVMs • Slab SVMs • Epsilon-SVR • Implement • Result
Single-Class SVM • Primary Object function • Concept
Single-Class SVM • Dual Problem • Decision function
Single-Class SVM • SVM TOY example
Outline • Introduction • Single-Class SVMs • Slab SVMs • Epsilon-SVR • Implement • Result
Slab SVM • Primary object function • Concept
Slab SVM • Lagrangian dual optimization problem • Decision function
Slab SVM • SVM TOY example
Outline • Introduction • Single-Class SVMs • Slab SVMs • Epsilon-SVR • Implement • Result
Epsilon-SVR • Primary object function • Concept
Epsilon-SVR • Lagrangian dual optimization problem • The approximate function
Outline • Introduction • Single-Class SVMs • Slab SVMs • Epsilon-SVR • Implement • Result
Implement • Single-Class SVM • Use LIBSVM directly • Slab-SVM • Modify epsilon-SVR in LIBSVM to slab-SVM • The result is parameter-sensitive • Modification: • decision function • The result is good but we don’t really know how to explain
Outline • Introduction • Single-Class SVMs • Slab SVMs • Epsilon-SVR • Implement • Result
Result(1) - One-Class SVM • Bunny • Original: 35947 • One-Class: 2810
Result(1) - Slab SVM • Bunny • Left: 8307。Right: 1013
Result(1) - Slab SVM • Bunny • Predict accuracy: 97.74%
Result(2) - One-Class SVM • Cactus • Original: 3337 • One-Class: 1617
Result(2) - Slab SVM • Cactus • Left: 996 • Right: 817
Result(2) - Slab SVM • Cactus • Predict accuracy: 96.9434%
Result(3) - One-Class SVM • Dinosaur • Original: 56194 • One-Class: 44963
Result(3) - Slab SVM • Dinosaur • Left: 13005。Right: 1018
Result(3) - Slab SVM • Dinosaur • Predict accuracy: 97.43%
Result(4) - One-Class SVM • Knot • Original: 10000 • One-Class: 8807
Result(4) - Slab SVM • Knot • Left: 1176。Right: 1036
Result(4) - Slab SVM • Knot • Predict accuracy: 96.9%
Result(5) - One-Class SVM • Screwdriver • Original: 27152 • One-Class:21728
Result(5) - Slab SVM • Screwdriver • Left: 6574 • Right: 884
Result(5) - Slab SVM • Screwdriver • Predict accuracy: 97.1752%
Result(6) - One-Class SVM • Rockerarm • Original: 40177 • One-Class: 4868
Result(6) - Slab SVM • Rockerarm • Left: 2090 • Right: 1029
Result(6) - Slab SVM • Rocker arm • Predict accuracy: 96.18%
Result(7) - One-Class SVM • Hole • Original: 4000 • One-Class: 3224
Result(7) - Slab SVM • Hole • Left: 648 • Right:1039
Result(7) - Slab SVM • Hole • Predict accuracy: 93.475%
Q&A Thank You