ANALYSIS OF GEOPOLYMER SLABS
USING
SUPPORT VECTOR MACHINE
Under the guidance of
R. Mourougane
Dept. of Civil Engineering,MSRIT.
Presented by
Naresh Reddy G.N
2nd Sem. M.Tech.
CONTENTS
1. Definition
2. Usage
3. History
4. Concept of SVM
5. Classification of SVM Models
6. Applications
7. Analysis of GPC Slabs
8. Limitations
9. References
Definition
Support Vector Machine (SVM) is a technique to analyze data and recognize patterns, used for classification and regression analysis.
Usage
Widely used for predictions and forecasting.
History
Originally developed by Vladimir N Vapnik in 1992.
Concept of SVM
H3 (green) doesn't separate the two classes. H1 (blue) does, with a small margin H2 (red) with the maximum margin.
Classification of SVM Models
Linear SVM Non Linear SVM
Basis of classification
Kernels: Mathematical functions such as linear, polynomial, sigmoid, radial basis function.
Applications of SVM
Text (and hypertext) categorization
Image classification
Face recognition
Bioinformatics (Protein classification, Cancer classification)
Hand-written character recognition
Engineering Applications (Environmental, Traffic data
analysis, Remote Sensing)
Versions AvailableSVMdark, SVMlight, SVMstruct, mySVM, libSVM.
Using SVMdark for comparison of experimental and predicted values for GPC Slabs.
Procedure
1.The experimental data is split into three sets (in the ratio 50%, 25%, 25%) namely training set, validation set and test set.
2.The train set and the validation set are used to fit in a suitable decision boundary based on regression analysis.
3.Further based on the constants obtained through optimizing the data, the test data is used to obtain the predicted values.
Input Output
Exp. Load (in KN)
Deflection (in mm)
Predicted Load (in KN)
Training Set(50%)
0 0
2 0.18
4 0.415
6 0.842
8 1.346
10 1.858
12 2.429
14 2.75
16 3.184
18 3.622
Validation Set(25%)
20 4.087
22 4.695
24 5.104
26 5,687
28 6.266
Test Set(25%)
30 6.786 ?
32 6.894 ?
34 6.936 ?
36 7.284 ?
38 7.316 ?
Slab Details
M60 Grade GPCSimply Supported
1080X500X65 mm
Input Output
Exp. Load (in KN)
Deflection (in mm)
Predicted Load (in KN)
Training Set(50%)
0 0
2 0.18
4 0.415
6 0.842
8 1.346
10 1.858
12 2.429
14 2.75
16 3.184
18 3.622
Validation Set(25%)
20 4.087
22 4.695
24 5.104
26 5,687
28 6.266
Test Set(25%)
30 6.786 29.701195
32 6.894 31.093738
34 6.936 33.241844
36 7.284 34.502888
38 7.316 36.617127
Limitations of SVM
Choice of the Kernel.
Selection of Kernel function parameters.
Depends only on a subset of the training data.
Speed
References
A User's Guide to Support Vector Machines.- Asa Ben-Hur & Jason Weston
The Solution Path of the Slab Support Vector Machine.
- Michael Eigensatz, Joachim Giesen & Madhusudan
Wikipedia
END