You are on page 1of 12

ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING

SUB CODE: 14EC803/A


Lectures: 4 Tutorial: 1 Practical: 0 Self Study: 0
Continuous Internal Assessment: 40 Semester End Examination (3 Hours): 60

UNIT-I
Artificial Intelligence (AI) - History And Foundation, AI Techniques, Problem Solving With AI
Models, Data Acquisition And Learning Aspects In AI.
Problem Solving – Problem Solving Process, Formulating Problems, Problem Types and
Characteristics, Problem Analysis and Representation, Performance Measuring, Problem Space
and Search, Toy and Real World Problems.
General Search Algorithms, Uninformed Search.
UNIT-II
Informed Search – Best First Search, Greedy Search, A* Search, AO* Search, Local Search
Algorithm and Optimization Problems.
Intelligent Agents: Rationality and Rational Agent, Performance Measure, Rationality and
Performance, Flexibility and Intelligent Agents, Types of Agents.
UNIT-III
Learning: What Is Machine Learning? , Learning Paradigms, Learning Concepts, Methods and
Models, Statistical Learning Methods, ANN Based Learning, Support Vector Machines (SVM),
Multi-Agent Based Learning, Distributed Learning, Adaptive Learning.
Learning Algorithms: Performance Matters, Active Learning, Learning Based on Limited
Information.
UNIT-IV
Neural Network Based Learning: History Development Of Neural Networks, Concepts And
Technologies of ANN, Feed-Forward NN, Feed-Back Networks, Pattern Associative Networks,
Competitive Learning, Self-Organizing Map, Network Design Issues.

TEXT BOOKS:
1. Artificial Intelligence: Building Intelligent Systems By Parag Kulkarni and Prachi Joshi, PHI
Publications.

REFERENCE:
1. Artificial Intelligence And Machine Learning By Vinod Chandra S.S And Anand Hareendran
S., PHI Publications.
2. Artificial Intelligence – A Modern Approach Second Edition, S.Russel And P.Norvig,
Pearson Education, 2003.
3. Tom Mitchell, Machine Learning, Mcgraw-Hill, 1997.

Course Objectives

Describe various techniques of Artificial intelligence(A.I) and analyze various aspects in


CO 1 identification of problem and problem solving methods
Demonstrate concept of solving complex problems through search and explain various types of
CO 2 intelligent agents with reference to complexity of problems
CO 3 Analyse various learning methods and acquire the knowledge about learning algorithms
CO 4 Apply neural network based learning methods in different applications.

ONE MARK QUESTIONS

UNIT-1

1. Write difference between semiotic models and statistical models.


a) Models which are based on sign processes are semiotic models and models refer to
representation and formalization of relationships through statistical techniques are statistical
models.
2. What are semiotic models?
a) Models which are based on sign processes are semiotic models
3. What are statistical models?
a) Models refer to representation and formalization of relationships through statistical techniques
are statistical models.
4. Define the term information
a) Information can be referred to as pattern underlying the data.
5. Define the term data.
a) Data refers to recorded facts.
6. Define the term data mining.
a) It refers as the extraction of meaningful information that is previously unknown and can be
useful potentially ahead.
7. What are the process that are included in data mining?
a) The mining process includes data cleaning, pre-processing, identifying and interpreting the
patterns, understanding the application and generating the target data with the consolidated
patterns.
8. What are well structured problems?
a) These problems yield a right answer when an appropriate algorithm is applied.
9. What are ill structured problems?
a) These are the problems which do not yield a particular answer.
10. What are linear problems?
a) It is the problem which can be solved or where decision can be obtained by linear solution.
11. Define the term non-linear problems.
a) It is the problem which cannot be solved or separated by linear equations.
12. What is intelligent agent?
a) It is the one which is flexible in terms of its action to get the desired outcome.
13. Define the term problem?
a) Problem is used in a situation when the desired objective is not obvious.
14. What is problem solving?
a) Problem solving is a process of generating solutions for a given situations.
15. Define the term Solution?
a) Solution to any problem is the collection of different states and set of operations.
16. What is meant by Deterministic problem?
a) This type of problem is also termed as single state problem. Each state is fully observable
and it goes to one definite state after any action.
17. What is meant by Non observable problem?
a) It comes under multi state problem i.e. application of operator can lead problem to multiple
states
18. What is meant by Non-Deterministic problem?
a) In this type of problem the effect of action is not clear. In every new state some new
information is added and then operator acts on the state.
19. Solution space is a tree structure of sates in which type of problem?
a) Non-deterministic or partially observable problem
20. Define unknown state space problems.
a) These are known as exploration problem in which states and actions are not known.
21. Define state space.
a) Solution to any problem is the collection of states and set of operations. This collection of
states is termed as state space.
Each of these states is achieved using the application of actions/operations to the previous state.
22. What is generality?
a) It should be able to handle all similar events irrespective of changing environment.
23. What is Transparency?
a) The reasoning behind the action taken should be visible to the user.
24. What is search?
a) Search is a general algorithm that helps in finding the path in the state space.
25. What is informed search?
a) It is a search that does not guarantee a solution but there is high probability of getting a
solution and makes use of heuristic.
26. What is uninformed search?
a) It is a strategy that generates all possible states in the state space and checks for the goal state.
Always find a solution if it exists.

27. Define state space method


a) During problem solving process an operator is applied to a state to move it to the next state
this goes on till final state is achieved. This approach of generating a solution is called state
space method.
28) What is problem solver?
a) .It is an algorithm that accepts problem description, domain description and boundary
conditions as input, and then searches for a solution.
29. What are the four ways to evaluate an algorithm performance?
a) Completeness, optimality, time complexity, space complexity
30. What are contours in state space?
a) Contours can be described as the boundaries between distinct regions with different properties.
31. What is the difference between informed and uninformed search
a) Informed search is a search that does not guarantee a solution whereas uninformed always find
a solution if it exists.
32. Define state space search.
a) It is the process of searching the state space for a solution to reach the goal.
33. What is an Agent?
a) An agent can be termed as an entity that can perceive the environment and act on it.
34. What are problem solving agents?
a) Problem solving agents are special type of goal based agents that formulates the goal as well
as the problem
35. Define control strategy.
a) It is a strategy that defines a way to follow while searching for the solution
36. What is forward search?
a) It is a strategy that proceeds from the initial state towards the solution.
37. What is backward search?
a) It is a strategy that proceeds from the goal state to the solvable initial state.
38. What is systematic search?
a) It is a blind search technique that is systematic in nature and is applied for small problem
space.
39. What are different types of systematic search?
a) Depth first search and Breadth first search
40. Define heuristic search.
a) It is a technique that improves the efficiency of search technique by providing guidelines.
41. What is the condition for optimality in search?
a) A search is optimal when the best solution is generated by the approach.
42. Define completeness.
a) Completeness defines that the algorithm will give the solution if there is one.
43 What is the significance of space and time complexity?
a) With space and time complexity we can address the memory required and the time factor in
terms of operations carried out.
44 Breadth first search follows ……………
a) Shallow node approach
45 Breadth first has ……………… structure
a) Queue data structure
46. Depth first search has ……………. Structure
a) Stack data structure
47 In BFS things happens in FIFO basis whereas in DFS LIFO
48 What is the idea of bi directional search?
a) In bi directional search, search is carried in both forward and backward directions.
49. What is depth limit search?
a) It is a version of depth first search where the expansion is limited to a specific depth limit.
50What is backtracking?
a) In DFS if we choose a path and could not find a solution we can think of going back at some
point and can find path to reach the goal state. This is called back tracking.
51. What is Artificial Intelligence?
AI is the simulation of human intelligence by machines that are programmed to think like
humans

Unit 2
1. Define heuristic function
a) A heuristic function is the one that guides the decision of selection of a path
2. What is Best first search?
a) It is a method that exploits DFS and BFS, switching between both to get the benefits of both
using heuristic
3. What is OR graph?
a) It is used to avoid revisiting of paths and also for propagation to the successor. It maintains
open and closed lists
4. Open list consists of……………..
a) List of nodes that have been generated and on whom the heuristic function has been applied
but yet not examined.
5. Closed list consists of………………
a) It contains the nodes that have already been examined.
6. Define the term admissible in search.
a) Any search algorithm is admissible if it always produces an optimal solution.
7. Define A* algorithm
a) It is a heuristic approach that performs search to compute optimal solutions.
8. What is consistent heuristic?
a) A heuristic h(n) is said to be consistent if for every node ‘n’ and every successor ‘s’ the
estimated cost to reach the goal state is not greater than the stepwise cost to reach ‘s’ added to
the cost of reaching the goal from ‘s’.
9. What is admissible heuristic?
a) A heuristic is said to be admissible if it does not over-estimate the cost.
10. What is AO* search?
a) It is a heuristic approach that maintains a graph rather than open and closed lists and performs
search.
11. Define local search.
a) Local search is a search that begins with initial solutions and iterates ahead exploring the
search space.
12. Which phenomenon is observed in hill climbing?
a) It uses the notion of heading towards a state which is better than the current state and moves to
the neighbour with a better score.
13. What is meant by local maximum?
a) This is a state better than the local region but not a global maximum since a better solution
exists, which is not in the vicinity of the present state.
14. Define the term plateau.
a) This refers to a flat area or space where neighbourhood states have the same value as the
present state and hence fails to determine the best direction to move on.
15. Define Ridge.
a) It is the search space at higher altitude than the surrounding that cannot be traversed by a
single move. It is a special kind of local maxima.
16. What is stochastic hill climbing?
a) It is a variant of hill climbing that selects at random among all possible uphill moves.
17. What is first choice hill climbing?
a) It is a variant of stochastic hill climbing where the successor is generated randomly until the
one generated is better than the current state.
18. What is random restart hill climbing?
a) It conducts a series of hill climbing searches from randomly generated initial states.
19. What is evolutionary hill climbing?
a) It is a genetic algorithm based search that performs random mutations and keeps the one that
are better.
20. Define simulated annealing.
a) It is an approach used for generating optimal solutions avoiding local optimas.
21. Write the significance of beam search.
a) It explores states or graph by expanding the most promising nodes in a limited set. At any
level it expands identified best node.
22. Is local beam search and stochastic beam search are same?
a) No, stochastic beam search is a variant of local beam search. Instead of selecting K successor
which are the best ones it selects K successor at random
23. Define Tabu search.
a) It is a search that tries non-improving move when gets stuck in suboptimal region. It maintains
a tabu list.
24. What are genetic algorithms?
a) Genetic algorithms are computational models based on biological evolution that are used to
solve optimisation problems.
25. What is optimisation?
a) Optimization is the process of improving a program's performance characteristics such as code
size (compactness) and execution speed.
26. Define reproduction.
a) It is the act of reproducing or making an exact copy of a potential solution that is done in
genetic approach.
27. What is crossover?
a) It is the act of swapping bit/gene values between two potential solutions, simulating the
mating of the two solutions that is used in genetic approach.
28. Define mutation
a) It is the act of randomly altering the bit/gene values in a potential solution’s that is used in
genetic approach.
29. What is an intelligent agent?
a) It is an entity that is autonomous in nature a good observant to detect the environment changes
with a capacity to govern its actions in timely fashion to achieve the goals.
30. What is meant by percept?
a) It is the window of the agent to the environment.
31. Define rational agent.
a) It is an agent that selects an action that would maximise the performance, based on percept
sequence and built in knowledge.
32. What are table driven agents?
a) It is based on simple table, where the entries are used to determine an action with respect to
percept.
33. What are simple reflex agents?
a) It is a simple rule based agent that takes into account only the current percept.
34. What are model-based reflex agents?
a) In these, the agent uses internal state, which depends on percept history is maintained and
reflects some of the unobserved aspects of the current state.
35. What are goal based agent?
a) It keeps building knowledge with reference to goal.
36. What are utility based agent?
a) On the basis of utility function, it maps the degree of happiness so that better decisions are
taken.
37. Define learning agent
a) These are the agents with an ability to learn and respond in unknown scenarios.
38. What is the role of sensor in agent-environment relationship?
a) A sensor allows the agent to sense environment and perceive the present state.
39. What is the role of actuator in agent-environment relationship?
a) An actuator allows the agent to take actions with reference to the environment perceived.
40. Which actions make an agent the most successful?
a) Rational actions are those which can make an agent the most successful.
41. List few examples of rational agents.
a) Rational agents can be a person, governing body, decision authority, firm, machine or
software.
42. Which terms describes the performance measure of an agent?
a) Performance measure is expressed in terms of efficiency, speed, solutions obtained, energy
consumed.
43. Why agents needs to be flexible?
a) Intelligence demands flexibility. Hence agents needs to be flexible.
44. What are the two important elements in learning agents?
a) Performance element and learning element
45. What is the role of learning element?
a) The learning element is responsible for making improvements in behaviour as it proceeds or
comes across more and more scenarios.
46. What is the role of performance element?
a) Performance element is responsible for the selection of external actions.
47. What are the different properties of intelligent agent?
a) 1.Responsive 2.Pro-active 3.Social
48. Define the term veracity
a) Intelligent agent should be truthful. It is not expected to hide information or lie.
49. When the intelligent agent is said to be pro-active?
a) If the intelligent agent exhibit opportunistic, goal directed behaviour and take the initiative
wherever necessary then it is said be pro-active
50 What is meant by social property of intelligent agent?
Intelligent agent should be able to interact with the other artificial agents, or humans in order to
compete problem solving.

UNIT -3

1. Define the term hyper plane


a) It is a line in SVM that can classify data
2. What is Support vector?
a) The vectors near the hyperplane are the support vectors.
3. Define Machine learning
a) Machine learning is building and exploring of methods for programming
computers to make them learn
4. List out few machine learning applications.
a) Social media, Marketing and sales and Health care Industry
5. What is active learning?
a) It is based on the outcomes of one query , desired outputs can be obtained for new
data
6. Active learning falls in which category of machine learning?
a) Semi-supervised learning
7. List the various types of machine learning algorithms.
1.Supervised learning 2.Un-Supervised learning 3.Reinforcement learning

8. Define absolute bias


a) Hypothesis is limited to definite set of functions.
9. Define preference bias.
a) Hypothesis that is minimal according to some ordering scheme across all
hypothesis is selected.
10. What is confusion matrix?
a) It represents the performance of any classifier.
11. List any two goals of machine learning
i.) Produce some learning algorithm with practical value
ii.)To facilitate in building intelligent system that can be used in solving real life
problems
12. List any two challenges of machine learning.
i.)Availability of limited learning data and unknown perspectives
ii)Accurate, compact and precise knowledge building is difficult and complex task
for the learning process in machines
13. What is labelled data?
a) Aware of the output for a specific set of inputs
14. What is unlabelled data?
a) The possible outcome is absent for specific set of inputs.
15. What is positive sample?
a) It provides information that it belongs to a particular category.
16. What is negative sample?
a) It provides information that it does not belongs to a particular category.
17. What is training set?
a) A set of positive as well as negative samples forms the training set.
18. Define error in back propagation algorithm.
a) Difference between actual output and the expected output is referred as error.
19. What is rote learning?
a) Rudimentary form of learning, which focusses on memorisation
20. Differentiate between classification and regression.
a) The output variable in regression is numerical whereas in classification it is
discrete
21. What is semi supervised learning?
a) When labelled and unlabelled data used for learning then it is called semi
supervised learning
22. What are true positive?
a) Data belonging to class are correctly predicted.
23. What are false negative?
a) Data have been accepted as a class but got rejected.
24. What are true negative?
a) Data not belonging to class are also correctly predicted.
25. What are false positive?
a) Data is classified in a particular group wrongly
26. What is Reinforcement learning?
a) In this type of learning, learner does not explicitly know input output sequence,
but gets reward or punishment from the environment as feedback.
27. Define ensemble learning?
a) It is the one where multiple learning algorithms are trained and ensemble of
hypothesis is selected from hypothesis space and their predictions are combined.
28. Define supervised learning?
a) It refers to learning under supervision where labelled data is used.
29.Define unsupervised learning
a) Data is not labelled but learning takes place based on generic properties like
similarity, closeness.
30. Which type of learning is suited for forecasting in financial market?
a) Unsupervised learning
UNIT-4
1. Define tree pruning?
a) Tree pruning reduces the size of decision trees by removing parts of the tree that do not provide power to
classify instances.
2. What is the role of bias in neural networks?
a) It is an optional entity that can have weight generally 1
3. What is back propagation?
a) It is an algorithm that adjusts the weights by considering the errors between the output and expected
values
4. What are self-organizing maps?
a) It is an unsupervised learning where the winner neuron shares the weight change with the
neighbours
5.Define competitive learning
a) It is an unsupervised learning where the winner neuron s allowed to change the weights
during the learning phase.
6. Write about minimum-error-rate in classification?
a) For any given problem, the minimum probability of error is
achieved by the LRT decision rule; this probability of error is called
the Bayes Error Rate or minimum error rate
7. What are Hopfield networks?
a) This is single layer fully connected network used for pattern association in auto
associative way.
8. Define activation function?
a) It specifies the output of the neuron and check whether neuron is active or it is in active.
9. Define heviside neuron
a) A neuron that makes use of simple step function
10. Write the significance of learning rate
a) It is a parameter that controls the weight adjustments.
11. What is ANN?
a) It is a network comprising many neurons that work in co-ordination to carry out certain
specific task
12. What is Multilayer Perceptron?
a) It is a feed forward neural network that has one or more hidden layers
13. What is recurrent networks?
a) This is a network where the connections between the nodes are in both the directions along
with loops.
14. What are pattern associative networks?
a) It is the one that generates output pattern itself
15. What are auto associative networks?
a) It performs unsupervised task where the correction of input pattern occurs.
16. List applications of SOM
a) Project prioritization and selection Seismic facies analysis for oil and gas exploration
Failure mode and effects analysis and Creation of artwork

6 Mark Questions

1. Explain the concept of Support Vector Machines.


2. State the learning paradigms of Machine Learning and explain each of them.
3. Write a note on:
i) Adaptive learning.
ii) Active learning
4. Explain how learning from observations takes place by giving examples
5. Discuss the need for semi-supervised learning and multi-perspective learning.

You might also like