CSE2AIF – Artificial Intelligence Fundamentals Exam Information and Preparation Advice
Examination date: Duration:
Total Marks:
Saturday 31 October 2020 4 hours
180
The exam will be available from 9AM and will be available for 12 hours from that time. Note that it will close completely at 9PM, so you should start before 5PM to allow enough time to complete the exam.
The exam is designed to be completed in 3 hours. The extra hour is to account for possible technical issues and to give you time to upload your solutions.
Once you start the exam, your individual 4-hour timer will begin to count down. You will have 4 hours from the time you start. You may start at any time in the 12-hour window.
The exam will be delivered using an LMS quiz. The exam will be separated into two parts. The first part will be questions which you can type your answers directly into the LMS; the second part will be answers which may require handwriting/drawing/etc. You will be required to scan and upload these solutions within the quiz as a PDF or DOC/DOCX file, before the 4-hour timer is finished. It is your responsibility to ensure the correct file is uploaded and that it can be opened and read by a human.
General Advice
• Do not get ‘stuck’ on a question. If a question proves more difficult to answer than you initially thought that it might be, then don’t waste time – move on to another question.
• Write clearly.
• Make sure you are familiar with scanning/photographing your solutions and
uploading them before the exam.
Academic integrity
The exam you complete must be individual work and you cannot seek assistance from other people during the exam. It is NOT permitted to collude (interact with a classmate) or consult any third party either electronically or in person (which is called contract cheating). These are both forms of academic misconduct and you may be subject to student discipline processes in the event of an act of academic misconduct. The policy can be found here: https://www.latrobe.edu.au/students/admin/academic-integrity
You will not be able to access the exam until you have verified the declaration of independent participation, which states that you will independently undertake the exam and agree to comply with all academic integrity requirements. To make this declaration, go to the “Exam” tile on the CSE2AIF LMS page, and click on the link, “Declaration of individual participation”. Please read carefully. You must read and understand this declaration, and agree to it.
Page 1
Content covered
The following is a list of the main topics covered in the unit. All topics are assessable.
• Lisp and Prolog programming
Through the course of the subject we covered basic Lisp and Prolog programming. You are expected to be able to write simple functions in Lisp. You are also expected to be able to write simple procedures in Prolog.
• State Space Search
o Formulatingstatespacesearchproblems o Searchstrategies
▪ Breadth-first search
▪ Depth-first search
▪ Depth-first search with depth limit
▪ Depth-first search with iterative deepening ▪ Best-first search
▪ A*-search (A-star)
o Propertiesofsearchstrategies ▪ completeness
▪ time complexity
▪ space complexity
▪ optimality/admissibility o Heuristicsforbest-firstsearch
▪ admissibility ▪ informedness
• Game Playing
o Minimax
o Minimaxtofixedply o Needforheuristics
o Alpha-betapruning
• Logic and Automated Reasoning
o Logicasaformalismforrepresentingknowledge o PropositionalLogic
▪ Truth tables
▪ Tautologies and contradictions
▪ Equivalences
▪ Inference rules
▪ Limitations of Propositional Logic
o PredicateCalculus(First-orderlogic)
▪ Existential and Universal Quantification
▪ Converting English sentences to Predicate Calculus ▪ Equivalences
o Inferencing
▪ Inference rules
▪ Soundness & Completeness
▪ Inferencing using forward-chaining
▪ Inferencing using backward-chaining ▪ Resolution Refutation
▪ Unification
Page 2
• Production Systems and Expert Systems
o Componentsofaproductionsystem
o Data-drivenvsgoal-drivensearch
o Controlofsearchthroughconflictresolution o ImportanceofProductionSystemsforAI
o Componentsofanexpertsystem(architecture) o Needforexplanationinexpertsystems
o Knowledgeacquisitionbottleneck
o Advantagesofexpertsystemapproach
• Knowledge Representation
o SemanticNetworkasumbrellatermforalargefamilyofgraph-based
knowledge representations
o Semanticnetworksasanalternativetopredicatecalculusforknowledge
representation o Frames
o AdvantagesanddisadvantagesofKR‘languages’(logic,rules,structured objects)
• Machine Learning
o DefinitionofInductiveLearning
o Terminology(examples,instances,attributes,features,etc) o DecisionTreesandtheID3algorithm
o Calculatingentropyandinformationgain
o Distinctionbetweensymbolicandsub-symbolicapproaches o FeedforwardNeuralnetworks(descriptivetreatmentonly) o Problemofoverfitting,andstrategiesforavoidingit.
Text Book References
References are for Luger, Artificial Intelligence, 5th and 6th editions
Lecture
References
Introduction to AI
5th Edn.: Chapter 1, Introduction to Part II 6th Edn.: Chapter 1, Introduction to Part II
State Space Search
5th Edn.: 3.0, 3.1, 3.2, 4.0, 4.1, 4.2, 4.3 6th Edn.: 3.0, 3.1, 3.2, 4.0, 4.1, 4.2, 4.3
Game playing
5thEdn.: 4.4,4.5 6thEdn.: 4.4,4.5
Logic
5th Edn.: 2.0, 2.1, 2.2, 2.3 6th Edn.: 2.0, 2.1, 2.2, 2.3
Resolution Refutation
5th Edn.: 13.0, 13.2, 6th Edn.: 14.0, 14.2,
Logic Programming and Prolog
5th Edn.: 13.3 6th Edn.: 14.3
Production Systems and Expert Systems
5th Edn.: 6.2, 8.0, 8.1, 8.2 6th Edn.: 6.2, 8.0, 8.1, 8.2
Knowledge Representation
5th Edn.: Intro. To Part III, 7.0, 7.1 6th Edn.: Intro. To Part III, 7.0, 7.1
Machine Learning
5th Edn.: 10.0, 10.1, 10.3, 11.0, 11.1, 11.2, 11.3 6th Edn.: 10.0, 10.1, 10.3, 11.0, 11.1, 11.2, 11.3
Page 3
Exam Preparation
You should revise all lectures notes and lab exercises, and read the text references indicated above. While all of the content listed above is assessable, you could use the following as a guide. You should be able to do the following:
• represent a problem as a state space search problem;
• given some search tree, identify the order of nodes examined using the various state
space search algorithms;
• compare the properties of the various state space search algorithms;
• trace the order of node expansion for A-star search using a given heuristic;
• discuss properties of heuristics; e.g., admissibility, informedness;
• use minimax to determine backed-up values for a game tree;
• explain the need for pruning in game trees;
• apply the alpha-beta pruning procedure to prune a game tree;
• construct truth tables for propositional logic expressions, and be able to demonstrate
equivalences;
• given some knowledge expressed in the natural language English, choose a set of
predicates appropriate for representation of this knowledge in predicate calculus;
• translate English sentences into predicate calculus;
• apply inference rules to infer new knowledge from a database of predicate calculus
expressions;
• construct proofs using resolution refutation;
• describe the components of a production system and explain the difference between
forward-chaining and backward-chaining expert systems;
• describe typical architecture for an expert system;
• given a set of rules, apply data-driven and goal-driven reasoning using these rules;
• describe what is meant by ‘conflict resolution’, and identify a number of alternative
conflict resolution strategies;
• represent simple knowledge bases using semantic networks and/or frame
representations
• explain what is meant by ‘inductive learning’, and distinguish between ‘symbolic’ and
‘sub-symbolic’ approaches;
• explain what is meant by ‘overfitting’;
• explain the ID3 decision tree induction algorithm, including the calculation of
entropy and information gain;
• explain the general operation of an MLP;
• describe the typical learning curves for an MLP;
• design MLP architecture appropriate for some given dataset, and
• write Lisp and Prolog code.
Page 4