Support Vector Machine

Shashank Shanu

a year ago

Support Vector Machine
Support Vector Machine
Most of already know one of the most prevailing and exciting supervised learning models with associated learning algorithms that analyses data and recognize patterns is Support Vector Machines (SVMs).
SVMs can be used for solving both regression and classification problems. However, it is mostly used in solving classification problems. SVMs were first introduced by B.E. Boser et al. in 1992 and has become popular due to huge success in handwritten digit recognition in 1994.
Before the emergence of Boosting Algorithms, for example, XGBoost and AdaBoost, SVMs had been commonly used.
If you want to have a consolidated foundation of Machine Learning algorithms, you should definitely know this algorithm. SVMs is a powerful algorithm, but the concepts behind are also not as complicated as you may be thinking.

What is Support Vector Machine?

SVM comes under a supervised machine learning algorithm. It can also be used for classifications as well as regression problems.
In the SVMs algorithm, we try to plot each data points as a point in n-dimensional space (where n is a number of features) with the value of each feature representing the value of a particular coordinate. Then, we perform classification by finding the hyper-plane that differentiates the two classes very well as shown in the below image.
Maximum Marginal Hyperplane (MMH)
Maximum Marginal Hyperplane (MMH)

Why we Need SVMs?

The Logistic Regression algorithm doesn’t care about whether the instances are close to the decision boundary or not. So, the decision boundary it tries to pick may not be optimal. As we can see from the above image, if a point is far from the decision boundary, we may be more confident about our predictions that it classifies right class. So, the aim of the optimal decision boundary should be able to maximize the distance between the decision boundary and all other instances. i.e., maximize the margins.
That’s why the Support Vector algorithm is important and we need it, most of the time as it produces better results.
The main aim of applying SVMs is to find the best line in two dimensions or the best hyperplane in more than two dimensions in order to help us separate our space into classes. The hyperplane (line) is found through the maximum margin, i.e., the maximum distance between data points of both classes.
We will be taking an example to understand it
Support Vector Machines
Imagine we are having the labelled training set are two classes of data points (two dimensions): Alice and Cinderella. As we can see to separate these two classes, there are so many possible options of hyperplanes that separate them correctly. As shown in the image below, we can achieve exactly the same result using different hyperplanes (L1, L2, L3). But, if we try to add new data points, the consequence of using various hyperplanes will be very different in terms of classifying new data point into the right group of class.
Support Vector Machines
Support Vector Machines
Now you may be thinking that then how can we decide a separating line for the classes and Which hyperplane shall we use? Let’s see the below image and try to understand it.
Support Vector Machiness
Support Vector Machiness
Some of the most important terminologies of SVMs are mentioned below-
  • Support Vector
  • Hyperplane
  • Margin
The vector points which are closest to the hyperplane are known as support vector points because only these two points are contributing to the result of the algorithm, and other points are not.
If a data point is not a support vector so removing that data points has no effect on the model. But if we delete the support vectors points it will then lead to change the position of the hyperplane which will affect the result.
Note: The dimension of the hyperplane will depend upon the number of features is present. If the number of input features is 2, then the hyperplane will be just a line and if the number of input features is 3, then the hyperplane becomes a 2D plane. It becomes difficult to imagine when the number of features becomes more than 3.
The distance of the vectors from the hyperplane is called the margin, which is a separation of a line to the closest class points. We would like to choose a hyperplane that maximizes the margin between classes. The image below shows what good margin and bad margin are.
good margin and bad margin
good margin and bad margin
The objective of the SVM algorithm is to divide the datasets into different classes and to find a maximum marginal hyperplane (MMH) which can be done as follows−
  • SVMs will generate hyperplanes iteratively that segregates the classes in the best possible way.
  • Then, it will choose the hyperplane that separates the classes correctly.

Kernel in SVMs

SVMs algorithm has a technique called the kernel trick. These are functions which take low dimensional input space and transform it into a higher-dimensional space, i.e., it converts not separable problem to separable problem. It is mostly useful in non-linear separation problems.
How to map lower dimension to a Higher Dimension?
You will understand it better with the below image.
map lower dimension to a Higher Dimension
map lower dimension to a Higher Dimension
map lower dimension to a Higher Dimension
map lower dimension to a Higher Dimension
Some of the most frequently used kernels are shown in the image-
Kernels
Kernels

A simple implementation of SVM in Python

# Import libraries

import pandas as pd
import numpy as np
from sklearn import svm, datasets
import matplotlib.pyplot as plt

# Loading Iris Dataset

iris_data = datasets.load_iris()

# Dependent and independent dataset.

X = iris_data.data[:, :2]
y = iris_data.target

#Now we will plot the SVM boundaries with original data 

x_min, x_max = X[:, 0].min() - 1, X[:, 0].max() + 1
y_min, y_max = X[:, 1].min() - 1, X[:, 1].max() + 1
h = (x_max / x_min)/100
xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))
X_plot = np.c_[xx.ravel(), yy.ravel()]

# Providing the value of regularization parameter 

C = 1.0

# Creating SVM classifier object

svc_classifier = svm.SVC(kernel='linear', C=C).fit(X, y)

#plotting svm with linear kernel

cf = svc_classifier.predict(X_plot)
cf = cf.reshape(xx.shape)
plt.figure(figsize=(15, 5))
plt.subplot(121)
plt.contourf(xx, yy, cf, cmap=plt.cm.tab10, alpha=0.3)
plt.scatter(X[:, 0], X[:, 1], c=y, cmap=plt.cm.Set1)
plt.xlabel('Sepal length')
plt.ylabel('Sepal width')
plt.xlim(xx.min(), xx.max())
plt.title('Support Vector Classifier with linear kernel')
Support vector classifier with linear kernel
Support vector classifier with linear kernel
# Import libraries

import pandas as pd
import numpy as np
from sklearn import svm, datasets
import matplotlib.pyplot as plt

# Loading Iris Dataset

iris_data = datasets.load_iris()

# Dependent and independent dataset.

X = iris_data.data[:, :2]
y = iris_data.target

#Now we will plot the SVM boundaries with original data 

x_min, x_max = X[:, 0].min() - 1, X[:, 0].max() + 1
y_min, y_max = X[:, 1].min() - 1, X[:, 1].max() + 1
h = (x_max / x_min)/100
xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))
X_plot = np.c_[xx.ravel(), yy.ravel()]

# Providing the value of regularization parameter 

C = 1.0

# Creating SVM classifier object

svc_classifier = svm.SVC(kernel='linear', C=C).fit(X, y)

#plotting svm with linear kernel

cf = svc_classifier.predict(X_plot)
cf = cf.reshape(xx.shape)
plt.figure(figsize=(15, 5))
plt.subplot(121)
plt.contourf(xx, yy, cf, cmap=plt.cm.tab10, alpha=0.3)
plt.scatter(X[:, 0], X[:, 1], c=y, cmap=plt.cm.Set1)
plt.xlabel('Sepal length')
plt.ylabel('Sepal width')
plt.xlim(xx.min(), xx.max())
plt.title('Support Vector Classifier with linear kernel')
Support vector classifier with rbf kernel
Support vector classifier with rbf kernel
I hope you enjoyed reading this article and finally, you came to know about the Support Vector Machine(SVM) algorithm, also you got to know how you can implement it using python.
For more such blogs/courses on data science, machine learning, artificial intelligence and emerging new technologies do visit us at InsideAIML.
Thanks for reading…
Happy Learning…

Submit Review

We're Online!

Chat now for any query