Definitions

Question 1

Symmetric Matrix

Answer 1

A square matrix (3x3, 2x2) is equal to it’s Transpose

Question 2

Feasible Region

Answer 2

Linear Programming, an area which satisfies two or more constraints in which all the solutions lie

Question 3

Feasible point

Answer 3

Linear Programming, a point within the feasible region

Question 4

Algorithm

Answer 4

A step by step procedure to be followed in calculations or other problem solving operations, particularly by a computer.

Question 5

Orthonormal Basis

Answer 5

Linearly independent set in which all vectors are unit vectors. (Vectors normalised to length 1). Orthogonal if dot product =0 0 .

Question 6

Local minimum

Answer 6

Smallest Value that a function takes at a point within a given area as opposed to the entire function domain.

Question 7

Examples of eigenvalues/vectors

Answer 7

PCA - Calculating optimal vector U, consider eigvec relating to largest eigval or covar matrix. EigenVecs represent variance of data in different directions. Used in Face Recognition.
Spectral Clustering - Weighted Adjacency matrix -> laplacian taken -> First k eigvecs or laplacian are used as input to k means clustering algorithm.

Question 8

Intelligent Algorithm Definition

Answer 8

Intelligence - Ability to vary state or actions in response to varying situations and past experience. IA satisfies this definition.

Question 9

How IA works

Answer 9

Step by step procedure giving unique output based on a situation. Supervised Learning - providing a training set of data allows it to adjust internal parameters. Gives it past experience which helps with new situations.
Unsupervised - no past experience, attempts to deduce some meaning in the data