Home Page > > Details

Help With COMP3308/3608Help With Matlab Assignment

 
 
COMP3308/3608 Introduction to Artificial Intelligenece (regular and 
advanced) 
semester 1, 2020 
 
Information about the exam 
 
 The exam will be online, via Canvas, un-proctored. It is set as a Quiz. 
 The Canvas site for the exam is different that the Canvas site we use during the 
semester. There are 2 exam sites: one for COMP3308 called “Final Exam for 
COMP3308” and another one for COMP3608 - “Final Exam for COMP3608”. The 
Exams Office will give you access to your exam site. 
 Students who are not in Australia and are in different time zones may apply for special 
arrangements for the exam: 
https://www.sydney.edu.au/students/special-consideration.html#time-zone 
 The duration of the exam is standard: 2 hours + 10 minutes reading time = 130 min. 
The Canvas Quiz will open at the scheduled time (as per your exam timetable) and close 
after 130 mins. 
 The exam is worth 60 marks ( =60% of your final mark). To pass the course you need 
at least 40% on the exam (i.e. 24 marks), regardless of what your mark during the 
semester is. 
 All material is examinable except week 1, week 13a (recommender systems), historical 
context, Matlab and Weka. 
 Tip: Have a calculator ready as there are some calculation questions. 
 There are 3 types of questions: 1) problem solving, 2) questions requiring short answers, 
and 3) multiple-choice questions (a small number). 
 
 
Sample exam questions 
 
In addition to these questions please also see on Canvas: 
Search: Quiz_Practice _with_Quokkas.pdf 
Bayesian networks: BN_practice_questions.pdf 
 
 
Question 1. In the tree below the step costs are shown along the edges and the h values are 
shown next to each node. The goal nodes are double-circled: F and D. 
 
 
 
Write down the order in which nodes are expanded using: 
a) Breadth-first search 
b) Depth-first search 
c) Uniform cost search 
d) Iterative deepening search 
e) Greedy search 
f) A* 
In case of ties, expand the nodes in alphabetical order. 
 
 
Question 2. Answer briefly and concisely: 
 
a) A* uses admissible heuristics. What happens if we use a non-admissible one? Is it still 
useful to use A* with a non-admissible heuristic? 
 
b) What is the advantage of choosing a dominant heuristic in A* search? 
 
c) What is the main advantage of hill climbing search over A* search? 
 
 
Question 3. Consider the following game in which the evaluation function values for player 
MAX are shown at the leaf nodes. MAX is the maximizing player and MIN is the minimizing 
player. The first player is MAX. 
 
 
 
a) What will be the backed-up value of the root node computed by the minimax 
algorithm? 
b) Which move should MAX choose based on the minimax algorithm – to node B, C or 
D? 
c) Assume that we now use the alpha-beta algorithm. List all branches that will be pruned, 
e.g. AB etc. Assume that the children are visited left-to-right (as usual). 
 
 
Question 4. Answer briefly and concisely: 
a) The 1R algorithm generates a set of rules. What do these rules test? 
 
b) Gain ratio is a modification of Gain used in decision trees. What is its advantage? 
 
c) Propose two strategies for dealing with missing attribute values in learning algorithms. 
 
d) Why do we need to normalize the attribute values in the k-nearest-neighbor algorithm? 
 
e) What is the main limitation of the perceptrons? 
 
f) Describe an early stopping method used in the backpropagation algorithm to prevent 
overfitting. 
 
g) The problem of finding a decision boundary in support vector machine can be 
formulated as an optimisation problem using Lagrange multipliers. What are we 
maximizing? 
 
h) In linear support vector machines, we use dot products both during training and 
during classification of a new example. What vectors are these products of? 
During training: 
During classification of a new example: 
 
 
Question 5. Consider the task of learning to classify mushrooms as safe or poisonous based 
on the following four features: stem = {short, long}, bell = {rounded, flat}, texture = {plain, 
spots, bumpy, ruffles} and number = {single, multiple}. 
 
The training data consists of the following 10 examples: 
 
Safe: 
 
 
Poisonous: 
 
 
a) Use Naïve Bayes to predict the class of the following new example: 
Show your calculations. 
 
b) How would 3-Nearest Neighbor using the Hamming distance classify the same example 
as above? 
 
c) Consider building a decision tree. Calculate the information gain for texture and number. 
Which one of these two features will be selected? 
 
You may use this table: 
x y -(x/y)*log2(x/y) x y -(x/y)*log2(x/y x y -(x/y)*log2(x/y x y -(x/y)*log2(x/y 
1 2 0.50 4 5 0.26 6 7 0.19 5 9 0.47 
1 3 0.53 1 6 0.43 1 8 0.38 7 9 0.28 
2 3 0.39 5 6 0.22 3 8 0.53 8 9 0.15 
1 4 0.5 1 7 0.40 5 8 0.42 1 10 0.33 
3 4 0.31 2 7 0.52 7 8 0.17 3 10 0.52 
1 5 0.46 3 7 0.52 1 9 0.35 7 10 0.36 
2 5 0.53 4 7 0.46 2 9 0.48 9 10 0.14 
3 5 0.44 5 7 0.35 4 9 0.52 
 
d) Consider a single perceptron for this task. What is the number of inputs? What is the 
dimensionality of the weight space? 
 
----- 
Note: The training data and new example will be given to you in a table, you are not expexted 
to derive them from pictures during the exam. 
 
 
 
The training data is: 
 
n stem bell texture number class 
1 short rounded spots single safe 
2 long flat ruffles single safe 
3 long flat ruffles multiple safe 
4 long rounded plain single safe 
5 long flat plain single safe 
6 short rounded plain single poisonous 
7 short flat plain single poisonous 
8 short rounded bumpy single poisonous 
9 long rounded spots single poisonous 
10 long rounded bumpy single poisonous 
 
The new example is: stem=long, bell=flat, texture=spots, number=single 
 
 
Question 6. Given the training data in the table below where credit history, debt, collateral 
and income are attributes and risk is the class, predict the class of the following new example 
using the 1R algorithm: credit history=unknown, debt=low, collateral=none, income=15-35K. 
Show your calculations. 
 
credit 
history 
debt collateral income risk 
bad high none 0-15k high 
unknown high none 15-35k high 
unknown low none 15-35k moderate 
unknown low none 0-15k high 
unknown low none over 35k low 
unknown low adequate over 35k low 
bad low none 0-15k high 
bad low adequate over 35k moderate 
good low none over 35k low 
good high adequate over 35k low 
good high none 0-15k high 
good high none 15-35k moderate 
good high none over 35k low 
bad high none 15-35k high 
 
 
Question 7. Use the k-means algorithm to cluster the following one dimensional examples into 
2 clusters: 2, 5, 10, 12, 3, 20, 31, 11, 24. Suppose that the initial seeds are 2 and 5. The 
convergence criterion is met when either there is no change between the clusters in two 
successive epochs or when the number of epochs has reached 5. 
 
Show the final clusters. How many epochs were needed for convergence? There is no need to 
show your calculations. 
 
 
 
Question 8. You task is to develop a computer program to rate chess board positions. You got 
an expert chess player to rate 100 different chessboards and then use this data to train a 
backpropagation neural network, using board features as the ones shown in the figure below. 
 
Select the correct answer (“Yes” or “No”) in the questions below. Select “Yes” for all issues 
that could, in principle, limit your ability to develop the best possible chess program using 
this method. Select “No” for all issues that could not. 
 
a) The backpropagation network may be susceptible to overfitting of the training data, since 
you tested its performance on the training data instead of using cross validation. 
 
b) The backpropagation neural network can only distinguish between boards that are 
completely good or completely bad. 
 
c) The backpropagation network implements gradient descent, so it may converge to a set of 
weights that is only a local minimum rather than the global minimum. 
 
d) You should have used higher learning rate and momentum to guarantee convergence to the 
global minimum. 
 
e) The topology of your neural net might not be adequate to capture the expertise of the human 
expert. 
 
 
Question 9. In the figure below, the circles are training examples and the squares are test 
examples, i.e. we are using the circles to predict the squares. Two algorithms are used: 1- 
Nearest Neighbour and 3-Nearest Neighbour. 
 
We are given the following results about the squares: 
 
Square Using 1-Nearest Neighbors Using 3-Nearest Neighbors 
1 - + 
2 - 
3 + 
4 + - 
5 - 
 
What will be the class of the following examples? Write +, - or U for cannot be determined. 
 
1) Circle L: 
2) Circle I: 
3) Circle H: 
4) Circle E: 
5) Circle K: 
6) Circle C: 
7) Square 6 using 1-Nearest Neighbour: 
8) Square 6 using 3-Nearest Neighbour: 
9) Square 3 using 1-Nearest Neighbour? 
10) Square 5 using 1-Nearest Neighbour? 
Contact Us - Email:99515681@qq.com    WeChat:codinghelp2
© 2014 www.asgnhelp.com
Programming Assignment Help!