Scribd
Upload
37 views2 pages

Derivation of The Margin

The document explains the derivation of the margin between two parallel hyperplanes defined by the equations W · X − b = −1 and W · X − b = 1. It details the computation of the perpendicular distance between these hyperplanes, concluding that the margin is given by r = 2/∥W∥, which indicates that the margin is inversely related to the norm of the normal vector W. The geometric interpretation highlights that the goal of Support Vector Machines (SVM) is to maximize this margin.

Uploaded by

praveencertificates11
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
37 views2 pages

Derivation of The Margin

The document explains the derivation of the margin between two parallel hyperplanes defined by the equations W · X − b = −1 and W · X − b = 1. It details the computation of the perpendicular distance between these hyperplanes, concluding that the margin is given by r = 2/∥W∥, which indicates that the margin is inversely related to the norm of the normal vector W. The geometric interpretation highlights that the goal of Support Vector Machines (SVM) is to maximize this margin.

Uploaded by

praveencertificates11
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

2

Derivation of the Margin ∥W ∥

Problem Setup
We are given two parallel hyperplanes:

W · X − b = −1 (First hyperplane),
W · X − b = 1 (Second hyperplane).

Here: - W = (w1 , w2 , . . . , wn ) is the normal vector, which is perpendicular to both


hyperplanes. - b is the bias term. - X = (x1 , x2 , . . . , xn ) represents points on the hyper-
planes.
Our goal is to compute the perpendicular distance (or margin) between these two
hyperplanes.

Perpendicular Distance
To compute the perpendicular distance between the hyperplanes, the following steps are
followed:

Normal Vector W
The vector W defines the orientation of the hyperplanes and is perpendicular to the
surfaces of both hyperplanes. The unit normal vector is given by:
W
,
∥W ∥
p
where ∥W ∥ = w12 + w22 + · · · + wn2 is the magnitude of W .

Point on the First Hyperplane


Let X0 be a point on the first hyperplane W · X − b = −1, so:

W · X0 − b = −1.

Moving Perpendicularly to the Second Hyperplane


To compute the distance to the second hyperplane, move along the direction of the unit
W
normal vector ∥W ∥
by a distance r. The new point becomes:

W
X0 + r · .
∥W ∥

This point lies on the second hyperplane W · X − b = 1.

1
Substitute into the Second Hyperplane Equation
W
Substitute X0 + r · ∥W ∥
into the equation of the second hyperplane:
 
W
W · X0 + r · − b = 1.
∥W ∥

Expand this equation:


W
W · X0 + r · W · − b = 1,
∥W ∥
∥W ∥2
W · X0 + r · − b = 1,
∥W ∥
W · X0 + r · ∥W ∥ − b = 1.

Use the First Hyperplane Equation


From the first hyperplane equation W · X0 − b = −1, substitute W · X0 − b = −1 into
the equation:

−1 + r · ∥W ∥ = 1.

Solve for r:

r · ∥W ∥ = 2,
2
r= .
∥W ∥

Conclusion
The perpendicular distance (or margin) between the two hyperplanes is:
2
r= .
∥W ∥

This shows that the margin depends inversely on the norm of the normal vector ∥W ∥.

Geometric Interpretation
- The hyperplanes are separated by the shortest distance along the direction of the normal
vector W . - The goal of SVM is to maximize the margin ∥W2 ∥ , which is equivalent to
minimizing ∥W ∥2 .

You might also like