Week 11 Introduction to Multivariate Analysis

Question 1

what assess the performance of a multi-variate classifier

Answer 1

type 1 and 2 errors

Question 2

what are the 3 types of learning

Answer 2

machine
supervised
unsupervised

Question 3

define machine learning

Answer 3

the automatic determination of the possible decision boundaries of a classifier

Question 4

define supervised learning

Answer 4

usage of a training dataset for which the true classification is know

Question 5

define unsupervised learning

Answer 5

performs classification without being instructed which characteristics to pick out

Question 6

what happens when training a multi-variate classifier

Answer 6

the configuration of the algorithm determines the number of degrees of freedom and through these the ability of the classifier to pick out small scale features of the training dataset

Question 7

how does the number of degrees of freedom alter a classifier

Answer 7

increasing the number of degrees of freedom of a classifier leads to a smaller bias but a larger variance

Question 8

what are the 4 multi-variate classification techniques

Answer 8

likelihood
k-nearest neighbour
artificial neural network
boosted decision trees

Question 9

what does the k-nearest neighbour approach do

Answer 9

automatically scales the size of the volume that is investigated with the density of entries

Question 10

what does the artificial neural network approach do

Answer 10

uses a combination of an arbitrary number of functions to pick out features in the dataset

Question 11

what is the artificial neural network equation

Answer 11

y(x) = ω0 + Σ[ωm*hm(x)]

Question 12

what are boosted decision trees

Answer 12

basically probability tree diagrams