Welcome! Log In Create A New Profile

Advanced

Sample exam paper

Posted by hammingweight 
Announcements Last Post
Announcement SoC Curricula 09/30/2017 01:08PM
Announcement Demarcation or scoping of examinations and assessment 02/13/2017 07:59AM
Announcement School of Computing Short Learning Programmes 11/24/2014 08:37AM
Announcement Unisa contact information 07/28/2011 01:28PM
Sample exam paper
October 19, 2009 10:57AM
There are two errors in the sample exam paper.

1. The sample solution for the 4-queens problem is not a solution. The queens in columns 2 and 3 are attacking each other.
2. The efficiency offered by alpha-beta pruning is significantly better than claimed. Assume that naive minimax needs to evaluate n states, it is claimed that alpha-beta pruning can reduce the number of states that need to be evaluated to n / 2. In fact, a-b pruning with n even can reduce the number of states that need to be evaluated to 2 * n^(1/2); i.e. approximately sqrt(n) << n / 2 for real-world scenarios (e.g. chess with a 4 ply or deeper lookahead).
Re: Sample exam paper
October 19, 2009 09:41PM
I also found two errors:

Question 2.2: There should be no line connecting (2,5) and (1,3,3) in the game tree because it is not a valid move
Question 4.2: The sentence ¬Tired(SK) v Hunt(SK) should rather read ¬Tired(SK) v ¬Hunt(SK)
Re: Sample exam paper
October 23, 2009 05:52PM
If I'm not mistaken, the sample paper is last year's exam paper. I haven't seen it yet, but the problems with the paper mentioned above seem to bring back memories.

If the paper is the final paper from last year (i'll confirm this a bit later), then take it with a bit of caution - there were many problems with the final exam paper, and if I remember correctly: in the end they awarded students automatically with 2 full questions' marks because of mistakes in the questions.
Re: Sample exam paper
October 25, 2009 11:35AM
There's a mistake in Appendix A as well. The information content values I for some values of p are inconsistent. If I(0.5+d) = x, then x = I(0.5-d) as well (from the definition of I). That would mean that I(0.6) = I(0.5+0.1) = I(0.5-0.1) = I(0.4). But, from the table I(0.6) = 0.98 and I(0.4) = 0.96. In fact the most accurate answer is that I(0.6) = I(0.4) = 0.97.

This is somewhat serious since one needs to use the values for p = 0.6 and p = 0.4.

Mistakes happen but the number of errors in this paper seems to be above what could be regarded as acceptable.

I trust that this year's exam paper has undergone more rigorous scrutiny than both the sample exam and the assignment questions which were also somewhat sloppily prepared. A pity since this is fun material.
Re: Sample exam paper
November 01, 2009 10:47AM
As noted above there is a problem with the solution 2.2 for the game of Nim (the transition (2,5) -> (1,3,3) is not a valid move)

Astonishingly there is another problem. The rows labelled MIN should be labelled MAX and consequently the procedure for choosing the best move actually chooses the worst. For example, the deepest node labelled (1,1,1,1,1,2) has a score of 0 implying a loss for the first player. In fact the node represents a win and in a conventiona/text bookl scoring system should be assigned the score 1. Compare the game tree in figure 6.1 on p.164 of the text book with the <sarcasm>solution</sarcasm> for the exam paper.

One could argue that perhaps the adversary is going to move first but that doesn't make much sense. One would do search when it is your move, you're not going to do the search until your opponent has moved. And in any case, if it is the opponent's turn to move, the opponent has a forceable win. Why assign the root node the score of 1 (when 0, or even more sensibly, -1 would be more appropriate).
Re: Sample exam paper
November 01, 2009 11:29AM
This paper seems absurd.

Can anybody explain in 6.3 how (Pheramone ^ ~StateBitOn) -> (Move ^ SetOn) from what we've been told?

Obviously the word "Pheramone" should be "Pheromone" but this spelling mistake is less serious than the questions. Seriously, who sent out this "solution"? Who set it? Who reviewed it?

My own qualifications to speak on AI are naturally limited but:
1 I have written several games that use alpha-beta pruning.
2. I have used simulated annealing as a local search technique to design antenna arrays and authored a widely cited IEEE conference paper which demonstrates the use of simulated annealing in array design (16 citations in journals and in the definitive book on the topic of local search for solving problems in electromagnetics).

Frankly I feel insulted by the lack of care that has gone into this.
Re: Sample exam paper
November 01, 2009 12:50PM
In case my previous post wasn';t sufficiently clear:
(Pheramone ^ ~StateBitOn) -> (Move ^ SetOn)
cannot be determined from the scanty information provided.

I.e., the "model answer" is, to put it charitably, nonsense.

In fact, it looks doubtful that any of the statements from propositional logic can be determined from what we've been told.

I trust that:
1. We will get some feedback from the staff concerned
2. This year's exam paper is free of errors.
Re: Sample exam paper
November 02, 2009 10:42PM
Although probably nearly too late to receive a reply, I would suggest emailing the lecturers as to this question.
Re: Sample exam paper
November 03, 2009 06:40AM
I did. In fact I also sent a request to the HoD asking him to investigate how this unprofessional piece of work could be sent out. I've never seen anything like it for the sheer number of errors. I trust that I'll get a response otherwise I will be escalating this..

Question 6.3, in particular, is inexcusably bad.

It's also worth noting that there were at least 3 errors in the assignments:
1. The multiple choice question that asked for the year that Deep Blue beat Kasparov.
2. The question that asked for the complete game tree for tictactoe (even though it's well-known amongst even amateur writers of two-player, zero-sum games that it's too big to draw).
3, A question that asked us to map the stupidly named predicates On and Clear to isOdd and >/ The question made no sense as the model answer made clear since the predicates were mapped to the contrary functions that we were told to map to!

This was:
1. An interesting course marred by
2. The sloppiest course material I have ever seen.
Re: Sample exam paper
November 04, 2009 07:47AM
I got a response from the lecturer concerned telling me that I'm welcome to ignore the paper. Assuming that's good advice (which I don't think it is): Don't try to make too much sense of stuff that is wrong. Unfortunately there's a lot of wrong stuff in this paper so perhaps throw the whole thing away. The quality assurance on this was deplorable.
Re: Sample exam paper
November 04, 2009 06:57PM
And another mistake found in the 4-Queens problem. Bu,t hey, the lecturer's advice is ":Ignore it!".

I must try the Unisa approach on our customers: "Oh, you found a bug? So? Just ignore it."
Anonymous User
Re: Sample exam paper
November 05, 2009 06:07AM
That's terrible attitude from a lecturer, its not like we have much to go on here in our exam prep. I must say this is one of the most poorly presented courses and I agree that the lecturers are sloppy
avatar Re: Sample exam paper
November 05, 2009 11:46AM
I've seen a number of modules that send out an errata note about mistakes in tutorial letters, then the next year they send exactly the same tutorial letter with the same errors followed by the same errata about those errors... *sigh*

---------
I'm Year '0110' Computer Student!
Sorry, only registered users may post in this forum.

Click here to login