In Support Vector Machine classification, what are the 'Support Vectors'?
Support vectors are the critical edge points that lie on the margin boundaries; removing non-support vector points does not alter the resulting decision boundary.
What is the 'Kernel Trick' in Support Vector Machines?
The kernel trick computes inner products in high-dimensional feature spaces where data becomes linearly separable, avoiding the computational expense of explicit coordinate transformations.
Which of the following functions are standard Mercer kernel functions used in non-linear SVMs?
Linear, Polynomial, and Radial Basis Function (RBF) kernels are all standard Mercer kernel functions applied in SVM classifiers.
Why is the K-Nearest Neighbors (KNN) algorithm formally classified as a 'Lazy Learner' (Instance-Based Learner)?
KNN is a lazy learner because training simply stores the instances. All computation—distance calculations and neighbor sorting—occurs when predicting new queries.
In K-Nearest Neighbors (KNN), how does choosing a very small value of k (e.g., k = 1) affect model behavior?
When k = 1, the classifier assigns each point to its immediate closest neighbor, capturing noise and creating irregular, overfitted decision boundaries.