I. Introduction
1. What is AI?
2. What Areas of AI does exist, and how do they relate to each other?
3. What Areas of Computer Science we learned about, and how do they relate to
each other?
4. Describe the agent-environment model
5. What is an <AI winter=?
6. Describe the Chinese room experiment
7. What point was the Chinese room experiment designed to demonstrate?
8. Give an example of a famous AI application
9. What is the technological singularity?
II. Problem solving
1. Classifying environments
a) What parts does an agent contain? (note: we discussed 2)
b) How would one define a task environment for an agent?
c) Given a task, give an example for all elements of PEAS
(Performance measure, Environment, Actuators, Sensors)
d) What were the environment properties we discussed? (note: there
were 7)
e) Give an example to all environment properties
1. Observability
2. Single/Multiple agent
3. Deterministic/Non-deterministic
4. Episodic/Non-episodic
5. Static/Dynamic
6. Discrete/Continuous
7. Known/Unknown
2. Decomposing problems
a) What steps were suggested for solving an AI problem? (note: there
were 8)
b) What properties would you consider when analyzing the data?
c) How would you divide a problem into sub-tasks?
d) How would you evaluate the different methods for sub-tasks?
III. Reasoning and retrieval
1. Search/Graph Search
a) What were the main types of search algorithms you met?
b) Compare the two uninformed search algorithms you learned (BFS
and DFS)
c) Describe the BFS/DFS search algorithm
d) Describe the A* search algorithm
e) What is a heuristic in the context of search?
2. Constraint satisfaction
a) What is an example of a constraint satisfaction problem (CSP)?
b) Describe CSPs (variables, domains, constraints).
c) What kind of constraints can we have? Describe them
d) How do Constraint Optimization problems (COPs) differ from CSPs?
e) What methods did you learn for the solution of COPs and CSPs
3. Game trees
a) What is needed from a game so that it could be analyzed using
trees? (list some examples)
b) Describe the elements of a problem to be analyzed using game
trees (game state, initial
state, terminal state, its utility, terminal test, players, actions, transition
models)
c) Describe the problem definition for a particular game
d) Describe a game using game trees (here, in some cases we would
give game to be described,
or we would ask you to choose a game you would want to describe)
e) What is a zero-sum game?
f) Describe the minimax algorithm. What are its
advantages/disadvantages?
g) Describe alpha-beta pruning
h) Demonstrate alpha-beta pruning on an example
4. Knowledge representation
a) You have learned about five types of knowledge. Can you recall and
explain them?
b) What knowledge representation techniques have you learned
about?
c) Describe frame representation. You can also use an example.
d) Describe semantic network representation (e.g., what tool we use for
representation, what do this tool represent). Use an example to help
illustrating it
e) What is ontological engineering? What are some of its limitations?
5. Logic, probability, fuzzy logic
a) What is a proposition? Give some examples of some special cases
b) Describe the truth table
c) What logic operators do you know of? Describe them
d) Given a logical formula, draw a truth table for it, and fill out its values
e) What properties are true for
negation/conjunction/disjunction/implication?
f) Explain the difference between propositional and predicate logic
g) What is Modus Ponens?
h) What rules of inference do you know of?
i) Explain the difference between deductive and inductive reasoning?
j) Explain the difference: logic agents, probability-based agents, and
utility-based agents.
k) What is the Principle of Maximum Expected Utility?
l) What is probability/conditional probability/prior probability?
m) What is the inclusion-exclusion principle/product-rule?
n) Describe Fuzzy logic in your own words. What is its utility?
IV. Machine Learning
1. Motivations and definitions
a) What applications of machine learning have you heard of?
b) What are the main steps of machine learning?
c) What key terms did you learn for machine learning? Explain these
terms
d) What areas of machine learning did you learn about? Give an
example for each
e) What classification methods did you learn?
f) Explain the difference between supervised and unsupervised
learning
g) Given a specific machine learning model (e.g. KNN, Random Forest,
etc.), or more general task (e.g. clustering, classification) answer
whether that particular model or task is supervised or unsupervised
learning
2. Nearest Neighbor, Decision Tree, Random Forest
a) Explain nearest neighbor classifiers. What property of features one
would have to be aware
of (think of pounds/kilograms, or centimeters/meters)
b) What distance measure can one use to find the nearest neighbors?
c) Explain the decision tree method
d) What is the connection between random forests and decision trees?
e) Explain the random forest classifier
f) What is reference to in the name of the random forest method
g) What advantages and disadvantages does the random forest
classifier has?
3. Linear models: linear regression, linear classification
a) What is regression?
b) Describe linear regression
c) What more advanced regression methods have you learned about?
d) Describe linear classification
4. Bayes classifier
a) Define the Bayes theorem for calculating conditional probability
b) What is the independence assumption?
c) How does the Naïve Bayes classifier work?
d) What are the advantages and disadvantages of the Naïve Bayes
classifier?
5. Neural Networks and Deep Learning
a) How artificial neurons are connected to biological neurons?
b) What is the role of the activation function, and what examples of it
did you learn?
c) How is a typical neural network built?
d) Explain Expectation Maximization (what are its steps, how do these
steps look)
e) How does gradient-descent work?
f) Describe Convolutional Neural Networks. How do typical CNN
architectures look?
g) What are Recurrent Neural Networks, and what variants have you
learned about?
6. Evaluation metrics and other practical tips
a) What metrics do you know for evaluating regression models?
b) What metrics do you know for evaluating classification models?
c) What is the confusion matrix? Explain its elements
d) Explain the difference between accuracy and F-score
V. Applications and ethics
1. Robotics
a) What are the fundamental questions of robotics?
b) When discussing me examples of properties one would control?
c) What is considered as the o f n a v i g a t i o n?
d) What factors are needed for path planning?
e) What algorithm can you implement in a robot in order to enable said
robot to get out of a simple maze?
f) What are some challenges of computer vision?
g) List applications of computer vision in robotics and explain one in
more detail
h) What types of robots did you learn about? (note: there were 6
categories)
i) Discuss some applications where robots can be helpful
2. Natural Language Processing
a) What is the difference between NLU and NLG? (what do these
abbreviations stand for?)
b) Give an example of NLU and NLG
c) What are the main units of the language
d) What is sentence segmentation?
e) What is tokenization?
f) What is the main difference between NLU and NLG, provide an
application for each one?
g) What are the three types of documents discussed for document
classification?
h) What is a bi-gram?
i) What is the bi-gram language model?
j) If we applied bi-gram on l be the outputs?
k) What can we filter out from the text in the process of filtration?
l) What is stemming/lemmatization? Provide an example.
3. Ethics in AI
a) Name some examples of technology shaping our perception (what
we see as important).
b) Can you recall some AI other sources)?
c) What four important elements of ethics were discussed in the AI
Ethics video?
d) What do these four elements entail?
e) What are the necessary elements of trustworthy AI according to the
EU?
f) What is explainable AI? Can you explain why it is important?
g) What is F.A.T in AI?
4. Business in AI
a) What is a business model?
b) What are some common business model challenges when
developing AI solutions?
c) In relation to business models, what does an ecosystem refer to?
It can also occur, that we use the example of a particular application
(e.g. chatbot, self-driving car, text
classifier trained for a certain task etc.), and ask some of the following
questions for that application:
- How would you start working with the task?
- Can you draw a potential pipeline for the task?
- What ethical concerns can you think of in relation to the task?
- What explainability issues can you think of in relation to the task?