• Home
  • Textbooks
  • Image processing, Analysis, and Machine Vision
  • Object Recognition

Image processing, Analysis, and Machine Vision

Milan Sonka, Václav Hlavác, Roger Boyle

Chapter 9

Object Recognition - all with Video Answers

Educators

Chapter Questions

Problem 1

Define the syntax and semantics of knowledge representation.

Check back soon!

Problem 2

Describe the following knowledge representations, giving for each one at least one example that is different from examples given in the text: Descriptions (features), Grammars, Predicate logic, Production rules, Fussy logic, Semantic nets, Frames (scripts).

Check back soon!

Problem 3

Define the following terms: Pattern, Class, Classifier, Feature space.

Check back soon!

Problem 4

Define the terms: Class identifier, Decision rule, Discrimination function.

Check back soon!

Problem 5

Explain the main concepts and derive a mathematical representation of the discrimination functions for:
(a) A minimum distance classifier
(b) A minimum error classifier

Check back soon!

Problem 6

What is a training set? How is it designed? What influences its desired size?

Check back soon!

Problem 7

Explain the principle of a Support Vector Machine (SVM) classifier.

Check back soon!

Problem 8

Explain why learning should be inductive and sequential.

Check back soon!

Problem 9

Describe the conceptual differences between supervised and unsupervised learning.

Check back soon!

Problem 10

Draw schematic diagrams of a feed-forward and Hopfield neural networks. Discuss their major architectural differences.

Check back soon!

Problem 11

For what is the back-propagation algorithm used? Explain its main steps.

Check back soon!

Problem 12

What is the reason for including the momentum constant in back-propagation learning?

Check back soon!

Problem 13

Explain the functionality of Kohonen neural networks. How can they be used for unsupervised pattern recognition?

Check back soon!

Problem 14

Explain how Hopfield networks can be used for pattern recognition.

Check back soon!
01:06

Problem 15

Define the following terms: Primitive, Alphabet, Description language, Grammar.

Mohamed Mohamed
Mohamed Mohamed
Numerade Educator

Problem 16

Describe the main steps of syntactic pattern recognition.

Check back soon!

Problem 17

Give a formal definition of a grammar.

Check back soon!
00:27

Problem 18

When are two grammars equivalent?

Ali Soave
Ali Soave
Numerade Educator

Problem 19

What is grammar inference? Give its block diagram.

Check back soon!
View

Problem 20

Formally define: A gruph, Graph isomorphism, Subgraph isomorphism, Double subgraph isomorphism.

Nick Johnson
Nick Johnson
Numerade Educator

Problem 21

Define Levenshtein distance. Explain its application to assessing string similarity.

Check back soon!

Problem 22

Explain why hill-climbing optimization approaches may converge to local instead of global optima.

Check back soon!

Problem 23

Explain the concept and functionality of genetic algorithm optimization. What are the roles of reproduction, crossover, and mutation in genetic algorithms?

Check back soon!

Problem 24

Explain the concept of optimization based on simulated annealing. What is the annealing schedule?

Check back soon!
03:09

Problem 25

List the advantages and disadvantages of genetic algorithms and simulated annealing compared to optimization approaches based on derivatives.

Jeremiah Mbaria
Jeremiah Mbaria
Numerade Educator
03:01

Problem 26

Define the terms:
(a) Fuzzy set
(b) Fuzzy membership function
(c) Minimum normal form of a fuzzy membership function
(d) Maximum normal form of a fuzzy membership function
(e) Fuzzy system
(f) Domain of a fuzzy set
(g) Hexige
(h) Linguistic variable

Lauren Shelton
Lauren Shelton
Numerade Educator
03:48

Problem 27

Use Zadeh's definitions to define formally: Fuzzy intersection, Fuzsy union, Fuzzy complement.

Aman Gupta
Aman Gupta
Numerade Educator

Problem 28

Explain fuzzy reasoning based on composition and de-fuzzification. Draw a block diagram of fuzzy reasoning.

Check back soon!

Problem 29

Explain the rationale behind using weak classifiers in the Adaboost training and classification process.

Check back soon!
00:59

Problem 30

The training process in boosting is sequential. Why is increasing weight placed on the training examples that were misclassified in the previous steps?

Rashmi Sinha
Rashmi Sinha
Numerade Educator

Problem 31

Provide a flowchart outlining the training process for a single tree of depth $D=3$ from a random forest. Using the flowchart, identify a path that a hypothetical single training pattern follows and describe how this training pattern contributes to the training process.

Check back soon!

Problem 32

Give an example of deriving a random forest posterior from posteriors of at least 3 individual trees:
(a) Using average-based ensemble model.
(b) Using multiplication-based ensemble model.

Check back soon!