Preparing for the COMP5318 Exams.

ticketdonkeyΤεχνίτη Νοημοσύνη και Ρομποτική

25 Νοε 2013 (πριν από 3 χρόνια και 4 μήνες)

108 εμφανίσεις




Preparing for the COMP5318 Exams.


You will be allowed to bring in one A4 page of double
-
sided notes.

The exam will have four questions. Each question will have three parts. The first part will be worth
15 pts. The remaining two parts will be 5 pts
each.

Question 1 (Bayes Theorem and Applications)


1(a) Will be about using Bayes Theorem for Classification. This will be very similar to the question in
the home work. The aim is to test whether you can completely solve the classification problem
using

Naïve Bayes.

1(b) This will also

be

application of Bayes Theorem but on a different task


for example clustering
or something else.

1(c) Will be the challenge question on Bayes Theorem. Slightly more difficult than 1(b).


Question 2 (Association Rule M
ining).


2(a) The first question will be on the Apriori type algorithm. For example, writing the pseudo
-
code
for the algorithm.

2(b) This will b
e on the complexity analysis of a data structure which is used in Association Rule
Mining. This could be the
Hash Tree or something similar.

2(c) This will be the challenge question. For example showing whether a measure is anti
-
monotonic
or not. Or some other property related to association rules.


Question 3 (Clustering)


3(a) Question on basics of K
-
means
clustering algorithm. In particular the iterative style of the K
-
means.

3(b) A question related to the fact that the K
-
means finds a local but not global optimal solution.

3(c) The challenge question. A question related to the EM Algorithm.



Question
4 (PCA/Maximum Likelihood)


4(a) A basic question on how to compute maximum likelihood estimates. Either for the Bernoulli,
Binomial or related distribution.


4(b) A

question on how to use Principal Component Analysis. Very similar to the tutorial on the 19
th

of October. It wil
l mostly be on interpretation of

PCA

results
rather than the mathematics.


4(c) A challenge question on PCA or Maximum Likelihood.