Visa resultat

Algorithms for machine learning and inference, Lp 3 Vt10, TDA231/DIT380

Your background

1. What is your motivation for taking the course?

2. Was your own background knowledge (in logic, combinatorics, probability) sufficient?

Yes»		12		42%
To some extent»		12		42%
No (I had to catch up many things)»		4		14%

Genomsnitt: 1.71

- Had knowledge in almost every concept talked in class. » (Yes)
- What do you mean by sufficient? Sufficient to have a geniuine discussion with Albert Einstein, or sufficient as a computer engineer? You might say sufficient to understand the course, and then I"ll tell you how deep you mean? I"d rather be required to know what I"d been taught, rather than random expectations. or in other words, please examinate what you teach.» (To some extent)
- In few of the things, I had to catch up. I had missed the first week of classes, that might have been the problem. Had some difficulty in the beginning.» (To some extent)
- Could have been better, but mostly sufficient.» (To some extent)
- My background in logic is good, but not in probability and combinatronics.» (No (I had to catch up many things))
- It was pretty tough keeping up, so a heads up on the website would be nice. Maybe there was one, I just didn"t see it.» (No (I had to catch up many things))
- Had not taken the course in combinatorics.» (No (I had to catch up many things))

Your own effort

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

28 svarande

At most 15 hours/week»		7		25%
Around 20 hours/week»		11		39%
Around 25 hours/week»		5		17%
Around 30 hours/week»		3		10%
At least 35 hours/week»		2		7%

Genomsnitt: 2.35

- The workload increases exponentionally in the term as you start the project not before half of the term...» (Around 20 hours/week)
- I spent much more time later on, during the project, than I did early in the study period.» (Around 20 hours/week)

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

28 svarande

0%»		0		0%
25%»		2		7%
50%»		6		21%
75%»		8		28%
100%»		12		42%

Genomsnitt: 4.07

- The reason for my low attendance was not the quality of the lectures, but rather the choice of lecture room. The E rooms were way too small, offered bad view for people in the back etc. If I have to stand or sit in the back between two people who never shower, I rather not go at all.» (25%)
- The lectures only provided a very very general overview. sometimes it sounded just like a summary of wikipedia. when you really want detailed knowledge about a topic you had to look for information by yourself. for every topic comming up with the same example is not the best idea as you can always figure out a know example by yourself but creating a new one is quite hard.» (50%)
- I had some overlaps with another course» (50%)
- I was absent the first week» (75%)
- I missed some classes because of some schedule problem.» (75%)
- Missed some lectures.» (75%)
- I just missed the first week, after that I was regular.» (75%)

Goals and goal fulfilment

5. How understandable are the course goals?

I have not seen/read the goals»		9		32%
The goals are difficult to understand»		2		7%
The goals give some guidance, but could be clearer»		12		42%
The goals clearly describe what I am supposed to learn»		5		17%

Genomsnitt: 2.46

- The goals towards the end of the course are not that clear, and let you expect a bit more than what is actually taught.» (The goals give some guidance, but could be clearer)
- the goals cover the course for some extend but not really a lot! The course gives a brought overview about some methods but _never_ goes into details. It is not mathematical at all!! Hard math was never covered and labeling a simple computation by the word "proof" does not make it mathematical! Reformulate the goals as otherwise people have wrong expectations of the course.» (The goals give some guidance, but could be clearer)
- I think some coding assignment would make the goal more clearer» (The goals give some guidance, but could be clearer)
- Even though we talked about many concepts we only scratch the surface.» (The goals give some guidance, but could be clearer)

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

21 svarande

No, the goals are set too low»		3		14%
Yes, the goals seem reasonable»		16		76%
No, the goals are set too high»		2		9%

Genomsnitt: 1.95

- The course felt more like a mere presentation of a few algorithms to "machine learn", and didn"t go into much detail like basing the choice of an algorithms on mathematical evaluation of the data. Also decision trees, and bayes rule in this form might be known from undergradute studies for most people...» (No, the goals are set too low)
- The goals do not fit to the course and let you expect much more» (No, the goals are set too low)

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

No, not at all»		4		17%
To some extent»		9		39%
Yes, definitely»		8		34%
I don"t know/have not been examined yet»		2		8%

Genomsnitt: 2.34

- the project could be about some small part of the course. one could have skipped all the lectures and copied the home assignments and write such a project and it would be fine.» (No, not at all)
- The laboratory is very unusefull, what is the point of clicking around in a program and getting some results if we dont know how the algorithms actually work. » (No, not at all)
- I made a project that tested one part of the course (if that). » (No, not at all)
- Project: actually is a report, and not a project ..» (To some extent)
- I don"t know, because I haven"t received grading results. What I didn"t like about the course was the grading criteria. I would have loved it more if I knew exactly what requirements I should fulfill, independent of the teacher"s preference.» (I don"t know/have not been examined yet)

Teaching and course administration

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

Small extent»		5		17%
Some extent»		9		32%
Large extent»		6		21%
Great extent»		8		28%

Genomsnitt: 2.6

- the lectures give a guidline what you should think about if you want to go deeper into a certain topic. it does not provide more information than wikipedia.» (Some extent)
- The lectures were too abstract, more focused for people coming from mathematics, or algorithms. So hard to follow most of the lectures.» (Some extent)
- The lectures were a nice second perspective (aside from the book) on things.» (Some extent)
- It could be better, if the teacher could use more visual depictions of the very abstract mathematical concepts he explained. That would be a great help. because I had really unfathomable lectures sometimes.» (Large extent)
- Peter was a great teacher, and the classes helped me a lot. » (Great extent)
- Lectures were great!» (Great extent)

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

Small extent»		2		7%
Some extent»		8		29%
Large extent»		11		40%
Great extent»		6		22%

Genomsnitt: 2.77

- i havent read a single page in the book. the lecture notes are more a summary of the content explained in the lectue than real notes.» (Some extent)
- The main literature (Mitchell) is not easy to read. For some topics i preferred to use Artificial Intelligence» (Some extent)
- I would have liked links to other webpages clarifying some of the concepts, as help to the lecture notes.» (Some extent)
- Lecture notes were very good. Course book was ok, but very boring.» (Large extent)
- the course pdfs online are great» (Large extent)
- The lecture notes are excellent.» (Large extent)
- The main coursebook was great.» (Great extent)

10. Did the lab assignment promote your understanding of the topics?

Yes, definitely»		5		19%
Yes, somewhat»		5		19%
No, but it was useful anyway»		4		15%
No, it was just another duty»		12		46%

Genomsnitt: 2.88

- there was no "lab assignment"...» (?)
- Yes. Yes. Yes. Application in practice. Loved it. especially the fact that I got extensive explanatory feedback from the course assistant, Azam, who helped to correct and overcome my shortcomings. The course exercises feedback was more of a error-finding correction. Feedbacks would just tell me I was wrong, and I haven"t understood the subject, or poining out some imperfections. I"d have preferred to have more interactive lectures, during which, I could ask WHY and get involved with the subjects, rather than sitting in a tightly-managed lecture time, and listen and not dare to distract the course of the teaching, in fear of, we don"t have time for that. let"s move on. More interactive lectures. Why not ask some people to simply explain the concept after teaching, to see whether they have understood the taught subject? instead of asking who couldn"t understand? no one would make him/her self embarrassed and say I didn"t!» (Yes, definitely)
- The lectures and the lasb are not well connected.» (Yes, somewhat)
- I can see how the Weka assignment was meant to give some practical hands-on experience, but for me it did not help my understanding, but was mostly a waste of time.» (Yes, somewhat)
- It was useful to get to know WEKA, but I would rather have seen a lab assignment such as the one in the algorithms course, so that we would have implemented some learning algorithm by ourselves.» (No, but it was useful anyway)
- Some of the questions were not really clear in the sensen of how exact the answer should be (e.g. find a lower bound, but lower bound was rejected saying it is not the "lowest possible" bound, etc...) .. beeing clearer here would have saved a lot of re-re-re-(re)*-submissions ... For some exercises it didn"t become clear what the learning outcome is (i do not consider "this algorithm isn"t suitable for this problem" a valuable lesson if it is already abvious from the beginning...» (No, it was just another duty)
- weka is one of the worst software packages i ever used. it crashes for random reasons, not only memory (on all of my platforms) and is not userfriedly at all. I cannot name you an alternative but you should look for one. The questions were more try and error what will be accepted and what not.» (No, it was just another duty)
- Useless» (No, it was just another duty)
- It was a total waste of time! Time that I desperatly needed for working on the machine learning project, working on resubmission of exercises and studying for exam in another course.» (No, it was just another duty)

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

Very badly»		0		0%
Rather badly»		1		3%
Rather well»		9		32%
Very well»		18		64%

Genomsnitt: 3.6

- Analog submissions are a pain :P» (Rather well)
- Why hand in everything on paper?» (Rather well)
- Even if it was hard to understand, it was really good the fact of getting feedback, it really helps to learn.» (Very well)
- Excellent course homepage and lecture notes.» (Very well)

Study climate

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

Very poor»		0		0%
Rather poor»		4		14%
Rather good»		5		17%
Very good»		15		53%
I did not seek help»		4		14%

Genomsnitt: 3.67

- Instead for the Lab exercise, which was well-feedbacked.» (Rather poor)
- Hard to find someone to ask. One way would be to book time for consultation, but I have been reluctant to do that.» (Rather poor)
- Some of the exercises were very hit and miss, it felt like trial and error, like guesswork, to get them done right. Sometimes I had no clue what was I doing wrong, because the email replies from the teacher were of no use...» (Rather good)
- Most answers could be answered using course material.» (I did not seek help)

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

Very poorly»		1		3%
Rather poorly»		0		0%
Rather well»		6		21%
Very well»		19		67%
I did not seek cooperation»		2		7%

Genomsnitt: 3.75

- We did our study group.» (Very well)

14. How was the course workload?

Too low»		0		0%
Low»		1		3%
Adequate»		14		50%
High»		10		35%
Too high»		3		10%

Genomsnitt: 3.53

- It took a lot of time to just understand, but overall it was good.» (Adequate)
- Difficult to handle both Weka assignment and a project (and exercises too). Maybe Weka+project could be rolled into one single project?» (High)
- 7 - very time demanding exercises, 1 time wasting supervised lab, 1 extremely time wasting (2 days) unsupervised lab when time was needed for project work, 1 big project to finish in 1,5 week, which is also the exam week, so we are assumed to have other exams to attend to that week. Yes, we could have started one week earlier with the project, but NO, because we had a big and time consuming exercise to hand in that week! » (Too high)
- Examination principles: Lab assignments and a exam: Fine A project course: Fine Home-problems during the course: Fine A home exam: Fine But! We had home-problems AND Lab assignments AND a project.: NOT Fine => Too much!!! » (Too high)

15. How was the total workload this study period?

Too low»		0		0%
Low»		1		3%
Adequate»		6		21%
High»		14		50%
Too high»		7		25%

Genomsnitt: 3.96

- but I had 3 courses» (High)
- Extreme! I"ve had several occasions of migraine because of the very high stress-level the last 3-4 weeks, and I"ve only been taking 1,5 courses at the same time and I have been working many, many extra hours beside the lectures from the first exercise, so it"s not because I have left everything for these last weeks!» (Too high)
- I am taking three courses plus a language course» (Too high)
- This was entirely my own fault...» (Too high)

Summarizing questions

16. What is your general impression of the course?

Poor»		5		17%
Fair»		4		14%
Adequate»		3		10%
Good»		11		39%
Excellent»		5		17%

Genomsnitt: 3.25

- I really expected more! This course is very very basic and only provides short spotlights on some topics. If you already have some background in statistics or ai you learn nothing new. » (Poor)
- What is the point? Should we learn to think in a certain way? No: We have to have the mathematical knowledge from before. Should we learn algorithms and tools to achieve certain goals? No: We have to figure out every thing for ourselves, and there is no clear goal in what machine learning is used for or what it could be used for in the future. So the course is more of an instruction to the students that they should already know everything there is to know about machine learning, and if they don"t, it is very hard to catch up because the teacher is assuming you know everything he in going to say.» (Poor)
- I was looking for practical applications, not mathematical theory without effective examples where machine learning is really doing its job.» (Poor)
- The course was very theoretical, few of the students will probably remember any of the mathematical proofs for very long. Furthermore I don"t think I would be able to implement a single algorithm that was taught without extensive further studies.» (Fair)
- I have learned very little from this course, compared to the many, many hours of work that I"ve put down. I still don"t understand the concept of machine learning. The only thing that I"ve learned from are 1-2 of the exercises and the project (I"ve mostly been improving my MATLAB-skills in the project, not much machine learning related improvements).» (Adequate)
- I had no real problems solving the exercises but still I cannot appreciate so much what I"ve learned in this course. I think there is a large gap between the mathematical foundations of ML that the course teaches and how it can be used, and I guess the lab is supposed to bridge this gap. But I think it must be done in another way. Try to connect theory and practise closer in the lectures.» (Good)

17. What should definitely be preserved to next year?

18. What should definitely be changed to next year?

19. Additional comments