Coffee in the cafeteria, and nothing to do, really. Which is sort of odd considering that there is still one exam ahead of me, adaptive systems tomorrow morning. But the presentation is ready, in fact has been since mid-afternoon yesterday. The networks are still not learning anything, at least not so obviously I could claim and prove it. So all I can do is go there and explain (again) what I have done, and what one might do, given a lot more time. ‘Cause sometimes I have that nagging thought that I was real, like real close to a breakthrough. But I have had that feeling quite often in the 190 hours I spent on this project. And while it defies my incomplete understanding of neural networks that this should just not work at all, in spite of all my attempts to make it, it’s possible.
190 hours. And a half. That’s, honestly, completely insane. Recall again the official requirements that fix the amount of study time, outside lecture and practicum, for any regular course as 132 hours over the course of the term, for an average student. Now I am certainly not average. I study a lot. But still, 132 hours times five courses is 660 hours. On top of 48 hours for attendance, times five, thus 240 hours. 900 hours total, divided by 15 term weeks, that’s a 60-hour week. Did I say this before? It’s insane.
Needless to say, even for me, actual study time has never been even close to these official figures. At least not over all courses in a term. I have been keeping a log of my study hours for three terms now, and almost in every case my actual hours fell way short of those provided for in the reguations. For instance I spent only 17 hours on business administration, part II (an oral presentation instead of an exam), and 45 hours on databases (a very limited and predictable exam). On the other hand, there have always been the programming and software engineering courses that ate a lot of time. Over the past three terms I have spent 176 hours on programming, part II; 122 on software engineering, part I; and a staggering 230 hours on part II. Though this figure is debatable as it includes over 50 hours of learning Git and Angular during the long term break–certainly related, necessary, and helpful, maybe even basic (particularly Git), but nevertheless somewhat extracurricular. I also invested most time in the various math courses: Over 110 hours each for automata theory and logic, around 140 apiece for graph theory and algorithms.
But 190 hours for a term project in a compulsory choice module (which means, an optional subject) certainly beats that all. Because this time was spent entirely during the term, and in fact most of it concentrated on a few weeks in late May and early June. Surely this is evidence of my maze and my neural networks having become an obsessive pet project for me. Even more the pity that it didn’t work out. Never have I invested such a large amount of time during such a short time frame with so little success to show for it!
So let’s hope we get at least a good grade in return for all that effort. From what I hear, most of my peers had little or no success with their adaptive systems project either, but at least they haven’t wasted a lot of time on them. (Wasting a lot of time seems to become my trademark anyway. In the transportation network simulation project for software engineering, part II, I wrote almost as many lines of code as my programming partner, ca. 3,000, but it took me about half as long again). There is no telling, yet, how the professor actually grades (as I write this, the first of us is waiting for his exam, today–mine is tomorrow), but he has the reputation of being rather critical.
Taken together, by the way, my log shows that this term, including the term break before it, I have studied a staggering 666 hours, a number however that includes 63 hours for our abortive participation in an academic research project and 15 hours on miscellaneous things such as preparing for the math tutorial course and speaking with our math professor about a possible bachelor thesis. And still, the remainder sums up to almost 600 hours! A good deal more than the 524 hours in the second term, when the cumulative stress nearly killed me and certainly contributed to driving me into a severe depression in the subsequent term break. Let alone the 476 hours I limited myself to, willy-nilly, in the third term. But then this term there was no really easy course or exam. Even intelligent systems, by no means complex in content, was so packed with widely divergent areas of knowledge it needed around 80 hours of preparation; and still I felt insecure before the exam. Not that more studying would have made any difference, seeing how we had two hours for an exam that were really two rolled into one.
So let’s see how I’ll do, time-wise, in the fifth term. For two of the five courses on the curriculum for that term will not be graded: A seminar, where you have to prepare a paper, and a project, where you program something in a small group. The effort for those two should be limited. Of the other two exams, certainly one (my compulsory choice module on process mining) will be oral, but probably two (information system architecture as well, if it will be taught by our software engineering professor, which is likely), and an oral examination is always more relaxed and less effort. So there is really only one regular, written-exam course, distributed systems. This sounds manageable, but then it better be, for I will be working two days a week. No idea if and how this will work out. Maybe I will have to postpone one or two courses, study seven terms instead of six. Then on the other hand, if I can just, somehow, survive the fifth term, the sixth will be a lot more doable, for apart from one course (IT security) and a third compulsory choice module it will be reserved for the bachelor thesis. And universal practice, if you work, is you try and do the thesis on a project at work, so that it doesn’t take extra time. In other words, then it will be just work and two courses, and that certainly should be doable. We’ll see.