Exponential Functions Since the Independent variable occurs in the Exponent...
To begin with it is imperative to understand that Support Vector Machine falls under Supervised Learning. That would mean that the Algorithm functions on Training Data, but with a twist.
The working of Support Vector Machine
This is a type of Algorithm that is predominantly applied to classify objects. Though the application of Support Vector Machines is vast, yet for the case of simplicity, we shall limit the discussion to introductory level alone.
Support Vector Machine, general understanding
Support Vector Machines can be best understood with the idea that the Algorithm is made up of three words giving us a literal meaning to the working of this Algorithm and they are, Support + Vector + Machines.
Expanding upon these terms we understand that the efficient functionality of the Machine depends upon the Support offered by data points in directing the decision of the Machine.
The word direction is what causes the rise of the term, Vector [by definition, Vector is a physical quantity that has Magnitude and Direction]
We live in a world where the heights of information overload has had a serious side effects on the mental cognition of people. This has erected challenges to administrators in various sectors.
With cross dressing and sex change all being permitted under the gratification of liberal approaches, has given rise to some serious issues to the smooth functioning of societal norms.
A 5 year old child walks into a washroom and a few minutes later a medium built muscular looking lady, follows the child into the washroom. To the unsuspecting eyes, all is normal and well. But No!
Minutes later, screams were heard which were caught by a vigilant janitor, who rushes into the washroom, and alarmingly she finds out that, in one of the enclosures the muscular looking lady to be in a compromising position with the child, with the child trying everything within physical limits to break free.
Though perturbed yet alert minded, the Janitor’s calls for immediate help and the muscular looking lady is apprehended. Upon examination, it was found that the perpetrator was in fact a man who has undergone sex change.
As reported, though the transsexual world may claim that cutting of one’s genitals is sufficient enough to change a man into a women, yet still based on the number of such incidents especially in relation to the crimes occurring in the Women’s room suggests otherwise.
Now with the dawning of AI, issues such as these can be curtailed. This is where Support Vector Machines comes into play. Just by merely installing a facial recognition surveillance Cam at the entrance of the women’s room, the incoming data object is captured and classified and appropriate actions taken.
Support Vector Machine, is a Supervised Learning Machine Algorithm, that looks at the data and determines its classification.
Let us suppose we have the Sample data which has Facial Geometry on the Y-axis and Shoulder width on the X- axis, this observation graphic model would help us to classify whether the data object is a Man or a Woman.
Women having more better leaner faces while men having broader shoulders. This is presumption on which the model works.
After suitably classifying the data objects it is time to draw the decision boundaries. Before we draw a decision boundary, you ought to know what is a decision boundary?
Decision boundary is a Straight line drawn on the void [empty space] space between the two Data classes. The Challenge is to draw the most appropriate decision boundary that would help in classifying a man from a woman.
If we do not have the optimal Decision boundary, we could incorrectly classify a man to be woman and let him pass through.
Below we shall illustrate two models and indicate their differences.
This model is not Optimal, the reason being, the distance between the male data point and the female data point is not maximum.
This model can qualify to be Optimal, however, in later posts we shall clearly explain the statistical calculation that goes into achieving the optimal hyperplane.
What are support vectors: These are data points that the margin pushes up against or data points that are closer to the opposing class.
In brevity, the Algorithm only considers the Support vectors, anything other training data can be ignored.
It is also imperative to understand from now on, we shall term the separating line as Hyperplane, owing to the varied dimension we shall encounter, however we a more intricate details will be explained in the later posts.
D+: the shortest distance to the closest positive point
D-: the shortest distance to the closest negative point
(D+) + (D-) = distance between the support vectors (also called as the Distance Margin)
By finding the Largest Distance Margin, we can get the Optimal hyperplane. From the Optimal hyperplane, we would easily be able to classify the incoming Data, as to which side it fits in.
Note: Ensure that the distance between the male data point and the female data point should be wide as possible.
0 responses on "Support Vector Machine"