Aktuella utvärderingar

Visa resultat

Här kan se resultatet från utvärderingen och exportera statistiken till ett annat program. Det går också att göra en enkel filtrering genom att klicka på svarsalternativen och kommentarerna eller en avancerad filtrering genom att använda knappen längst ned.

Artificial Neural Networks, FFR 135, FFR135

Status: Avslutad
Öppen för svar: 2009-10-15 - 2009-12-15
Antal svar: 28
Procent av deltagarna som svarat: ?%
Kontaktperson: Bernhard Mehlig»

Your own effort

1. How many hours per week did you spend on this course?

We mean total time, that is, it comprises the time you spent in class and the time you spent on your own work. Try to estimate the average time over the entire study period.

28 svarande

At most 15 hours/week»2 7%
Around 20 hours/week»8 28%
Around 25 hours/week»7 25%
Around 30 hours/week»9 32%
At least 35 hours/week»2 7%

Genomsnitt: 3.03

- I did the course last year, so I didn"t need to spend that much time this year.» (At most 15 hours/week)
- It"s very hard to tell, more of a guess from my part» (Around 25 hours/week)
- A bit above average perhaps.» (Around 25 hours/week)
- A big amount of work but very interesting since we can organize our work as we want. I discovered a new way of working (not the same in my home country : 30 hours of courses each week but a small amount of work at home). Maybe precisions about the quality of the reports we are expected to do should be given. Because the more we want to write interesting reports, the more time it takes.» (Around 30 hours/week)
- The workload was pretty heavy with regards to the number of credits. In my home university, such a workload is worth around twice as much credits. With regards to the learning outcome, however, the workload is fair. I learned a lot of things, half of which are unrelated to ANNs things I learnt indirectly.» (Around 30 hours/week)
- Writing the programs, and debugging, was quite time consuming. » (At least 35 hours/week)

2. How large part of the teaching offered did you attend?

28 svarande

0%»0 0%
25%»1 3%
50%»5 17%
75%»9 32%
100%»13 46%

Genomsnitt: 4.21

- I did the course last year, so I didn"t need to spend that much time this year.» (25%)
- The morning times are far from ideal. Maybe the times should be more balanced.» (50%)
- 50% due to collisions with other course. Class was overall excellent.» (50%)
- Very pleasant to wake up in the morning to go to this course ! » (100%)
- or rather 90 percent» (100%)
- I missed 2 lectures because I the other course I followed had a lecture at the same time. However my partner went to them and we exchanged the content of both lectures afterwards. » (100%)

Goals and goal fulfilment

The course syllabus states the course goals in terms of learning outcomes, i.e., knowledge, skills and attitudes to be acquired by the student during the course.

3. How understandable are the course goals?

28 svarande

I have not seen/read the goals»5 17%
The goals are difficult to understand»0 0%
The goals give some guidance, but could be clearer»7 25%
The goals clearly describe what I am supposed to learn»16 57%

Genomsnitt: 3.21

- It is at the very last course that I understood the whole outline of the course. Maybe a recall of where we are in the outline could be done (just during a few minutes).» (The goals give some guidance, but could be clearer)
- The goal are quite clear, but the means are not. I think the whole principle of example sheets should have been made clear since the course description. » (The goals give some guidance, but could be clearer)
- I didn"t read them until now, and it"s my own fault. I had no idea that the course was that focused on report writing. I have put most of my efforts into programming and understanding the concepts. I would"ve put more time into the writing.» (The goals clearly describe what I am supposed to learn)
- As usual (as for all courses) the goals are not understandable before I have taken the course, afterwards however they are very understandable (might imply that the goals have been reached)» (The goals clearly describe what I am supposed to learn)
- Since the subject is a bit tricky when you"re new to it, this question becomes somewhat difficult to answer.» (The goals clearly describe what I am supposed to learn)

4. Are the goals reasonable considering your background and the number of credits?

Answer this this question and the succeeding one, only if you do know the course goals.

24 svarande

No, the goals are set too low»0 0%
Yes, the goals seem reasonable»23 95%
No, the goals are set too high»1 4%

Genomsnitt: 2.04

- This course represents a lot of credits so it is normal that we work a lot on the topic.» (Yes, the goals seem reasonable)
- As I said before, the workload is too high with regards to the credits. Mostly because I misunderstood what you expected of us for the example sheets, see below.» (No, the goals are set too high)

5. Did the examination assess whether you have reached the goals?

23 svarande

No, not at all»1 4%
To some extent»7 30%
Yes, definitely»14 60%
I don"t know/have not been examined yet»1 4%

Genomsnitt: 2.65

- The examination primarily assesses if one can write an error-free computer program. Which is a good thing to be able to do, but not a primary goal.» (To some extent)
- Having a smart friend helps you a lot more than putting in extra effort into the course, maybe change to just fail pass and possibly work in pairs» (To some extent)
- Yes because we have to construct networks and if the results are not those expected I know that I misunderstood something.» (Yes, definitely)
- The example sheets are excellent for learning and understanding the course elements!» (Yes, definitely)
- This has been perhaps the best examination I have had so far (engineering physics chalmers)» (Yes, definitely)
- I think one cannot pass all the example sheets without a decent minimum of understanding. But the process of passing the example sheets could be made more painless.» (Yes, definitely)

Teaching and course administration

6. Comparing take-home vs regular exam do you:

28 svarande

Much prefer regular exam»0 0%
Prefer regular exam»3 10%
Have no opinion»2 7%
Prefer take-home exam»11 39%
Much prefer take-home exam»12 42%

Genomsnitt: 4.14

- A combination with also a written exam, where the parts of the course that are not covered by the take-home exams, also are tested. » (Prefer regular exam)
- I am fed up with those silly exams where you have to learn by heart all the course and even if you didn"t understand something, you can have a good mark. Here, by taking our time, at home, by choosing how to organize our work, we are in the situation we will be when we will have a job, or do a PhD. » (Much prefer take-home exam)
- Written assignment is much better than regular exam for almost anything except things where you do calculations by hand..» (Much prefer take-home exam)
- I do not like the regular exam situation and always perform worse during these. Furthermore it is not as stimulating to study towards a written exam as it is towards a take-home exam which you can really dive into. Keep it the way it is! (and persuade others to do the same)» (Much prefer take-home exam)
- Learning for life.» (Much prefer take-home exam)
- Regular exams are the worst way to assess a student"s level. One cannot make a projection of a student on a segment of the real axis. A student is a too many dimensional entity. take-home exams allow for some kind of trial and error, some discussion between the teacher and the student that give a better idea of the student, even if the final grade is still only a 1 dimensional projection. In my humble opinion, the best way to teach and evaluate is an as-full-as-possible-scale project, in small groups (2 to 4) with no final exam (the grading is made as time goes by) » (Much prefer take-home exam)

7. To what extent has the teaching been of help for your learning?

28 svarande

Small extent»1 3%
Some extent»3 10%
Large extent»16 57%
Great extent»8 28%

Genomsnitt: 3.1

- Easy to follow. Easy to understand.» (Some extent)
- In some cases the teaching was great (I especially appreciated all the diagrams and visualizations on the black board), but too often Bernhard will get entangled in lengthy calculations where students could quickly loose track of what we actually are doing. In some instances these calculations were only marginally important, and in some very important. I would advise that some time is spent, and the importance of the calculation is better described before the calculations are initiated.» (Some extent)
- Good lectures complementing the notes.» (Large extent)
- Inspiring mostly, more than technical knowledge.» (Large extent)
- The question is what to learn. The lectures were not so connected to the encountered problems when solving the example sheets, and vice verse. Generally, though, the lecture subjects, like stability analysis, are more interesting than the problems of software construction. » (Large extent)
- I learn the theory with the teacher and the rest I learnt was with the exemple sheets.» (Great extent)
- The example sheets teached me every thing else I learnt in this course that is not directly related to ANNs and that demanded so much work.» (Great extent)

8. To what extent has the course literature and other material been of help for your learning?

28 svarande

Small extent»3 10%
Some extent»11 39%
Large extent»8 28%
Great extent»6 21%

Genomsnitt: 2.6

- I didn"t read the book proposed. I just use at the beginning wikipedia. The course was enough because it was very exhaustive.» (Small extent)
- The lecture notes on their own is not covering everything. Did not buy haykin, maybe should have.» (Some extent)
- The lecture notes were good to jog memory and remind about the lecture, but if one had missed the lecture, it was almost impossible to decipher the lecture notes. Given the one book is out of print, and other only loosely follows the course, it becomes essential to go to the lectures.» (Some extent)
- I"m sorry to say that, because I hate when people tell it to me, but your writings on the course material is almost unreadable. Typeset it in LaTeX if you find the time, or ask one of the lab"s slave, oops phd student, to do it for you.» (Some extent)
- lecture notes» (Large extent)
- The literature was too expensive for me, but I eventually found it on the internet (not the same as holding a book in ones hands though). The thing with multiple sources is that they differ somewhat and it gets confusing. But it has worked.» (Great extent)
- The lecture notes are great. I rarely read course books and haven"t done so now either, but the lecture notes are great, they contain the essence of the important stuff.» (Great extent)
- Haykens was the material that I used the most for to solve the example sheet problems. » (Great extent)

9. How well did the course administration, web page, handouts etc work?

28 svarande

Very badly»0 0%
Rather badly»5 17%
Rather well»13 46%
Very well»10 35%

Genomsnitt: 3.17

- maybe use a webbased submission system» (Rather badly)
- I would"ve liked information about penalties for missing deadlines. Sometimes you almost have the complete program, but there"s some bug that you need a day more to find, because you started late. The question is then what to do, hand in on time or later? 2-weeks is very reasonable, but there usually are other courses that require computer time and programming that compete about time.» (Rather well)
- An improvement would be to update the "last update date" text on the course website when there is news on the website. This text still says 2 september 2009, but this course evaluation was not on the website that day. I also thought that there was going to be a document on the website with more information on the homeproblems, about what to hand in and things like that. Haven"t seen it.» (Rather well)
- Not entirely clear about goals and results of example sheets (ie max points). There were some information given exclusively in class which is not good.» (Rather well)
- Nothing to report.» (Very well)

Study climate

10. How were the opportunities for asking questions and getting help?

28 svarande

Very poor»1 3%
Rather poor»2 7%
Rather good»9 32%
Very good»11 39%
I did not seek help»5 17%

Genomsnitt: 3.6

- Please provide Kristian"s email address or room location on the web page - it was hard to locate him when questions arose with the excercizes. » (Rather poor)
- The teacher was helpful, but we only had three exercise classes, and no programming class.» (Rather good)
- Funny to see you get out very quickly to get your morning coffe, but you were available and often helpful, except for the "see with kristian" part, see remarks about example sheets below.» (Rather good)
- The break is very convenient because the teacher has time to answer questions. We don"t have that in my school and at the end of the course, the teacher or the students are always in a hurry and the questions stay without answer.» (Very good)
- I always feel welcome to ask questions» (Very good)
- One computerized-demo session would have been good.» (Very good)

11. How well has cooperation between you and your fellow students worked?

28 svarande

Very poorly»2 7%
Rather poorly»5 17%
Rather well»6 21%
Very well»10 35%
I did not seek cooperation»5 17%

Genomsnitt: 3.39

- I came to Sweden with another student of my class and we were used to work together and to help each other. Maybe it is shame not to have met others students.» (Very poorly)
- 2h per week with a lab assistant meant for working wouldn"t hurt» (Rather poorly)
- Some more effort could be spent in the beginning of the class to integrate different groupings (CAS students, other master students, etc) as they do not know each other since before. » (Rather well)
- I work with my old friends from engineering physics chalmers.» (Very well)
- Comparing my results with one of my fellow student more than once allowed me to find bugs.» (Very well)

12. How was the course workload?

28 svarande

Too low»0 0%
Low»0 0%
Adequate»14 50%
High»13 46%
Too high»1 3%

Genomsnitt: 3.53

- Good level» (Adequate)
- Felt like a lot of work, but was doable, which means it was about right :-)» (Adequate)
- High but the tasks were very pleasant to do.» (High)
- The workload was high for me because I am not so used to programming and made subtle errors here and there. The mathematics however was easy, so the course was not hard conceptually.» (High)
- High but adequate.» (High)
- But fair» (High)
- a bit uneaven between exercises and example sheets» (High)
- Your lack of precision of what was expected of us in the example sheet increased the workload to a quite high extend. A bit much for what it was worth (but don"t worry, it was worth a lot). When hearing you the first time you talk about the example sheets, it seemed that there were small scale scientific papers, and would be graded accordingly. After the first example sheet has been given back to us, and after that kristian talked about the second one, I understood that there were graded as normal exercises. Which was quite frustrating with regards to the amount of work I put in these first two example sheets. My advice for next year : either choose the exercise way, tell your student to spend as little time as possible on those and be far more precise about the answer you expect (What graphs, precise answers to precise questions) or choose the small scale scientific paper way, and ask your student for a bit of creativity and inventivity, ask open question and let the student decide how to answer it. But if you do it, grade accordingly : read the code, take it into account, take the apparent motivation of the student into account, take the style into account. This year, you were in the first for the grading but in the second for the explainations. Which is frustrating. » (High)
- Depending on the programming experience, the workload becomes quite different among the students. More "theoretical" examples would even out the load, I think. » (Too high)

13. How was the total workload this study period?

28 svarande

Too low»0 0%
Low»0 0%
Adequate»10 35%
High»15 53%
Too high»3 10%

Genomsnitt: 3.75

- In CAS absolutely adequate, on the low side. But that gives opportunity to take an extra course which is nice.» (Adequate)
- Or more like unbalanced. Not much to do in the first weeks and then there were suddenly relatively much to do (I"m a CAS-student).» (Adequate)
- Same as above :-) It was more than I expected it to be, most of my weekends and nights went into working with the handins.» (Adequate)
- The other course I took had a by-heart examination, the kind of material you learn the day before and forget the day after. It compensated for the workload of yours.» (Adequate)
- I was luky because I hadn"t not a lot of problem with the mathematics used and the other course I took didn"t require a lot of work (maybe I"ll have problem at the exam ^^ fingers crossed)» (High)
- I can say it was high if I compare to the other master programmes like applied physics and engineering mathematics. It seems Fundamental has a higher workload than cas though. But it wasn"t too high, I feel the work is needed to get a grasp of the courses.» (High)
- High but adequate.» (High)
- It was high but not too high. This kind of studies should be though» (High)
- The other course I took was Discrete event control and optimization.» (High)
- im lazy» (High)
- Both courses had a high workload, which made it the total workload to high.» (Too high)

Summarizing questions

14. What is your general impression of the course?

28 svarande

Poor»1 3%
Fair»0 0%
Adequate»2 7%
Good»12 42%
Excellent»13 46%

Genomsnitt: 4.28

- Too much math, too little practical tips on how to do neural nets for my liking.» (Adequate)
- Good as if somewhat too technical.» (Good)
- I discovered a new field of reaserch. The teacher seem to be very happy to teach and it is pleasant to follow the course.» (Excellent)
- Interesting topic well presented. Somewhere in the course is a transition from neural networks to the kohonen and other clustering algorithms. This could be done clearer as it can be quite confusing. No biggie though.» (Excellent)
- Very interesting and fun.» (Excellent)
- Or rather "very good".» (Excellent)
- You did not waste my time, and to my great disillusionment, only a few teachers can say so. Most teachers misuse PowerPoint, Impress @ Co, keep using the blackboard. Also keep being funny. I especially liked the parking tickets. » (Excellent)

15. What should definitely be preserved to next year?

- The enthusiasm and the general approach to the subject, from the physics (and not the mathematics) angle.»
- The teacher !»
- The very inspiring lecturer!»
- Focus on problem solving, briefing on motivations and mathematics - good compromise»
- Global summary of previous lesson»
- Good example sheets in general»
- Bernhard»
- Kohonen was fun, and stimulating. Very visual.»
- The pedagogical approaches taken.»
- The lecturer»
- Kohonens networks! That was the definitive highlight of the class for me.»
- Theacher, Literature»
- "question time lecutre"»
- The lectures style.»

16. What should definitely be changed to next year?

- Theory is important, but I would"ve liked getting a basic feeling of the course contents and algorithms before getting involved into theory. It is much easier to relate if you have some practical experiences from implementing some algorithm. Then thoughts and questions arise naturally, why does this work, and when? In my opinion, it would be more suitable to push theory a little bit later into the course (after some implementation part). There should also be clear information about how handins are graded. The feedback is not enough. Gustafsson does not seem to read code (which is understandable), but many explanatory comments or questions are usually written there. Sometimes you get hefty deductions saying "theres an error in your code", but without any references. For such a claim to be made I would like to know where it is. If the claim is based on bad results there should be a clear explanation. Some "errors" are there on purpose, and explained as interpretations in the code. Things can be implemented in different ways, which generate slightly different results. It can also have to do with different interpretations of certain tasks/functions. Also, people who use rather low-level languages are at huge disadvantage compared to those who make matlab-programs. Generally it"s easier to code in matlab, which gives more time for bug-detection, report writing and result evaluation. Writing non-matlab code is good for oneself, but puts you at a disadvantage looking at the grades (=time).»
- More information on the course page. What will be taught at which lesson etc.»
- The redaction of the exemple sheets : you shouldn"t write so long paragraphs, just put carriage returns, it will be ok. Either precise exaclty what you expect in the report, or leave more freedom in our writting and experiments but adapt the mark. »
- All relevant information regarding example sheets on course homepage (where, when, in what format and what the grading is)»
- Some of the lecture notes do not explain the full concepts of the unsupervised networks, needed for sheet number 4»
- If there is a requirement or important note that is directly related to an example sheet it would be of great help, if it was actually in the example sheet and not only mentioned during lectures and extra lectures.»
- Add some more examples classes»
- More on unsupervised learning»
- Nothing at all, to the setup of the course (examination etc.), this is the best course I have taken during my university studies»
- More help with programming. We did get ANY technical help with it, basically.»
- Add an informal meeting with last years students after half of the course to discuss the course and program contents.»
- More exercise sessions.»
- Handpick important proofs and calculations, collect other somewhere where the interested students can check them. Spend more time developing the intuition about different ANN, in which way they differ and in which way they are alike. Give more real life examples, both of the NN in living organisms and in applications in the industry. Give feeling about the current front-edge research in the NN - this is still a developing area. Some more history of the networks and people behind would be interesting, but probably hard to push time-wize. »
- The bulk of the examples. Better would be to also have a written exam on the more theoretical parts of the course. »
- Group works should be included.»
- some symbols were quite hard to distinguish in particular when handwritten which could get confusing (e.g. xi, zeta also teacher"s 0"s often looks like sigma)»
- Choose your way for the example sheets, wuthout ambiguity (see above).»
- The question oppertunities should be improved and there should be some programming help»

17. Additional comments

- Very nice course !»
- An enjoyable course!»
- I like that the mathematical theory behind the algorithms were teached on the lecture. However, I didn"t really had so much use of some of the mathematical theory. In some example sheets you could, in principle just implement the algorithm and don"t care about the mathematics. So I"d like some home problems where you have to work with some of the mathematics used in this course. Also I"d like to work more with unsupervised learning in the course.»
- i liked this course i think it"s good in its present condition»
- As mentioned before pass/fail might be better for the course considering the most beneficial thing for this course besides knowing it really well is having a friend to ask and that will not show how well the student understands the subject. Maybe add a little bit more focus on implementation e.g. efficiency or tips for a good implementation»
- Some things could be bettered, in stochastic order : -Readable course material. -(a bit) Less calculation, (a bit) more explaination »
- It"s very poor feedback on your assignments, it"s not ok to just get the comment "something seems to be wrong"!»

Kursutvärderingssystem från