Again in that thoroughly deflated state of mind when one exam is over and I can’t yet face up to the next one being just around the corner. And today’s intelligent systems exam was one of the most frustrating exams ever–in fact I think it was only ever topped by the Java programming exam in the second term in which even the best programmers among us didn’t manage to complete half of the assignments.
Today’s exam were really two exams–one by each of the co-teachers– rolled into one, but with the allotted time just about enough to complete one of them comfortably. I was jittery before it even started. After 20 minutes, when I realized I had done just one of the eight problems in the first part, I had my first panick attack. The second was when I realized, after 40 minutes, that I had absolutely no idea how to solve the Prolog programming problem–it was way over my head. And I am arguably one of the best Prolog programmers in the semester group!
I then completed the second part, which was mainly small questions aimed at testing our understanding of machine learning algorithms, but involved a naive Bayes classifier problem, which is a lot of manual computation and which I got wrong the first time and had to redo, wasting further precious minutes. With 30 minutes left, I returned to the Prolog program and decided, for lack of a better idea and above all, for lack of time, to sketch something very brief that with a lot of good will (which I am not sure the professor will muster) might look like something that might have a chance to work, though I am certain it does not. Then I returned to part II and wrote a few wild guesses to the questions I did not understand, such as why are neural networks not used for safety-relevant applications, but decision trees are.
With 3 minutes to go I realized to my horror that in the assignment that asked us to search a tree with six different algorithms I had simply overlooked one and done only five. I hastily scribbled the sixth, unfortunately one of the more involved, hill-climbing with backtracking, and then the professor collected the solutions, or what we hoped would pass as such, and we were done.
Not since the nightmarish Java programming exam in the second term have I had such a bad start into an exam period. My best guess is that if the professor grades the exam kindly, I can hope for a B, but really I can’t say, because I never had the time to double-check my frantically scribbled answers. If I made a lot of careless mistakes, it might be a C or worse. Well, maybe not worse.
And mind you, I consider such an outcome rather undeserved. For I had really studied a lot to prepare for this exam. Had the Prolog problem been in the general difficulty range of the professor’s previous exams (those that she did not do together with a co-teacher who added almost a full exam of his own!), I am confident I would have done fine, more or less. But as it was, with this impossible assignment looming over my head, I was hectic most of the time, which is a bad state to be in in any exam. Again I wonder if, with a lot less preparation and instead a more relaxed state of mind I could have done a lot better. But it’s of course equally possible that with less preparation I would have been panicky all the time and would not have achieved anything.
Oh well, now for getting this disaster out of my mind and concentrating on the next exam, computer networks on Monday, which should be more manageable, and above all predictable. Though after today I am no longer sure.