Question

15 pts Question 6 Describe in a few sentences what each of these algorithms learn from training data, and how they classify unseen data - (1) SVMs 2) Learned Decision Trees

Answer 1

1. SVMs :- In support vector machine, given the set of training data, the aim is to learn the hyperplane which will separate the training data with one side of hyperplane contains the data belongs to one class and another side of hyperplane contains data belongs to another class. So if the data is linearly separable then SVM will find the support vector whose distance from the training data of either class is maximized. Also in case the data is not linearly separable, then we define the penalty for misclassified data.

2. Learned decision tree :- Decision tree is used to classify the data to particular class by filtering the data based on certain attribute value. Decision tree use the attribute at particular node for the decision purpose on which filtering the data will give maximum information gain where information gain is calculating by checking the decrease in entropy when we filter the data based on value of attribute. Thus by traversing the decision tree from root to leaf node, we get the classification of data.

Please comment for any clarification.