SUL Topic 4 - SVM Flashcards

1
Q

Support Vector Machines (SVM)

A

Classification technique for predicting binary outcomes using labelled and balanced datasets

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Black box technique

A

Characteristic of SVM where individual variable influence cannot be determined

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Hyperplane

A

What SVM learns to classify data into two classes

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Kernel Trick

A

Method used to transfer input space to feature space when data points are not linearly separable

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Kernel function

A

Maps pairs of data points onto their inner products and transforms data from lower to higher dimensions

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Max-margin hyperplane

A

Result of mathematical minimization in feature space, identifying non-linear boundary in original data space

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Support vectors

A

Extreme data points that SVM focuses on for precise classification

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Linear and non-linear separable data

A

Types of datasets SVM can handle using kernel functions

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

High feature-to-sample ratios

A

A challenge faced by SVM requiring careful parameter tuning

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

SVM applications

A

Fields including medical imaging
Financial predictions
Pattern recognition

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Categorical attribute handling in SVM

A

Convert categorical values to numeric values

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Multi-class SVM

A

Combines two-class SVMs to handle non-binary classification

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Training time for SVM

A

May be very long for large-scale problems

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Accuracy Measure

A

Classification measure used to evaluate SVM performance

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Confusion Matrix

A

Tool used to assess SVM classification results

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

ROC Curve

A

Necessary result for comparing SVM method against others

17
Q

SVM flexibility

A

Ability to use custom kernels and combine functions for complex hyperplane creation

18
Q

SVM efficiency

A

Works well even with very small training sizes

19
Q

SVM limitation

A

Only performs binary classification, Not non-binary

20
Q

Data representation in SVM

A

Balls represent the dataset, with colors indicating different classes