SVM with kernel and code

1. Support Vector Machines (SVM)
and Kernels - A Beginner’s Guide
• Presented by: Rushikesh
• Date: February 2025

2. Introduction to SVM
• Support Vector Machine (SVM) is a supervised
machine learning algorithm used for
classification and regression tasks.
• Real-life analogy: Separating fruits based on
size and color.

3. Why Use SVM?
• Advantages of SVM:
• - Effective in high-dimensional spaces
• - Works well with clear margin of separation
• - Robust to overfitting
• Example: Classifying emails as spam or not
spam.

4. How SVM Works - Intuition
• SVM finds the optimal hyperplane that best
separates the data into classes.
• Margin: The distance between the hyperplane
and the nearest data point from each class.

5. Mathematical Understanding
(Basic Level)
• Margin = 2 / ||w|| (where w is the weight
vector)
• Goal: Maximize the margin to improve
classification.

6. Support Vectors
• Support vectors are the data points closest to
the decision boundary.
• They are critical in defining the position and
orientation of the hyperplane.

7. Linear vs. Non-Linear Data
• Linear data: Can be separated by a straight
line.
• Non-linear data: Requires transformation to a
higher dimension.

8. Introduction to Kernels
• Kernels help SVM handle non-linear data by
transforming it into higher dimensions.
• Analogy: Flattening a crumpled paper to
separate points easily.

9. Types of Kernels
• 1. Linear Kernel: Suitable for linearly separable
data.
• 2. Polynomial Kernel: Handles complex
boundaries.
• 3. RBF Kernel: Maps data into infinite
dimensions.
• 4. Sigmoid Kernel: Similar to neural network
activation.

10. Visualization of Kernels
• Different kernels project data differently to
enable separation.

11. Practical Applications of SVM
• Applications include:
• - Face detection
• - Text classification
• - Handwriting recognition
• - Medical diagnosis

12. Advantages and Disadvantages of
SVM
• Advantages:
• - Effective in high dimensions
• - Versatile with kernel trick
• Disadvantages:
• - High training time for large datasets
• - Not suitable for overlapping classes

13. Simple Code Snippet
• Python code using scikit-learn:
• from sklearn import datasets, svm
• iris = datasets.load_iris()
• clf = svm.SVC(kernel='linear')
• clf.fit(iris.data, iris.target)

14. Summary and Key Takeaways
• - SVM is powerful for classification tasks.
• - Kernels extend SVM to handle non-linear
data.
• - Practical and widely used in real-world
applications.

15. Q&A Session
• Feel free to ask questions!