This was a mini-course - just 8 classroom sessions - that I taught for the first time at PGSS in Summer 2001. The concept was to introduce students to a new research area of computer science, but without requiring any previous background. (Indeed, the students didn't have to touch a computer throughout the course.)
aspect abstain 1 2 3 4 5 avg assignments - 0 1 4 16 9 4.10 class sessions - 0 1 9 10 10 3.30 handouts 0 0 0 3 11 16 4.43 TA help 6 0 1 11 8 4 3.63
mean 2.243 hours stddev 1.096 median 2 hours range 1.0-5.0 hours midrange 1.5-2.5 hours
0 You went too fast for me regularly. 19 Sometimes a bit fast, but usually OK. 11 Sometimes a bit slow, but usually OK. 0 I regularly got bored waiting for you to move on already.
topic 1 2 3 4 5 avg Data mining 2 13 11 4 0 2.57 Neural networks 1 1 21 7 0 3.13 Reinforcement learning 0 4 9 15 2 3.50 Artificial life 0 6 18 6 0 3.00
On a scale of 1 to 5, how interesting were the class topics? (1 being ``boring,'' 5 being ``fascinating'')
topic 1 2 3 4 5 avg Data mining 1 12 9 6 2 2.87 Neural networks 0 2 3 16 9 4.07 Reinforcement learning 0 4 12 11 3 3.43 Artificial life 0 4 1 12 13 4.13
type of material 1 2 3 4 5 avg techniques / algorithms 0 6 9 11 4 3.43 research results 2 9 11 6 2 2.90 theory behind techniques 0 3 13 11 3 3.47 field applications 1 1 10 13 5 3.67
I would have liked to do more complex algorithms like linear regression using software packages such as [??], and maybe even other analyses such as least-squares and Chi-square. I would like to have covered less theory behind the techniques. I would not hav echanged the assignments - they were the perfect difficulty level, allowing us to practice and learn the techniques while not spending too many hours doing very hard problems. I enjoyed this class very much, and I hope that it will return as an elective next year, because it introduces statistical techniques in an interesting and enjoyable way (thanks to you, CBurch!)
I think we should actually use a computer to stimulate [sic] a lot of things we learn (reinforcement learning and artificial life). Moreover, I think there should be some class work to keep us awake. :)
I would have liked to spend more time on artificial life. Along with that, I would like to know where AI is how and where it is likely to go. For the assignments, I could have used some varied examples in class to further my understanding. Overall, I would have liked to see these techniques implemented on actual machines. This implementation would have showed exactly how efficient these techniques are on machines versus by hand. Overall, I loved this course and learned a lot! Thanks CBurch!!
I thought some parts of some assignments were repetitive, but I don't how you could change it. Is there any way you could demonstrate actual machine learning programs? I thought this was the most intriguing course I took. The fact I feel asleep some times was not your fault.
Overall the class was very interesting. It was definitely overwhelming at first, especially all the computations. I would have liked to spend a little more time discussing the techniques, the theory and research behind them, than dissecting the algorithms so much. Sometimes the textbook and class presentation seemed contradictory/confusing, but the TAs were a great help. A great class overall. CBurch rules!!
A little too much on algorithms, I would have enjoyed some discussions of the applications and purposes for Machine learning. But it was all new info, which was very interesting. Thanks for a lovely class.
The class was pretty interesting and challenging. The work was hard fro me because I had never done computer clasess before.
The formulas given in the packet were difficult to interpret at best. Although assignments weren't usually that difficult, they took a long time to write up all the iterations, which were usually very similar, and one mistake could make everything else wrong. Spreading out days on which different electives' homework is due would be nice. I really enjoyed the slides.
I thought the clas was very interesting, it was nothing like I could have learned in high school. I also liked the animated overheads. CBurch, you da man!
I greatly enjoyed this class - Machine Learning was my favorite elective. It was so neat to see how data is processed and how conclusions are reached from a baseline set of data. I never understood how computers reached conclusions and handled the data given them, but I feel that I have a good grasp of these ideas now. The homework was very reasonable and someone like me who has never really worked with computers before was able to handle it. I think that the format of the class was great, with enough time spent on each topic and with the homework reinforcing the concepts well. Thanks for making the class interesting and for doing a great job teaching!
The assignments were mostly cool except the reinforcement learning one. The chart was a little confusing and I couldn't figure out the packet. Otherwise the packets were very helpful! I liked the artificial life stuff we covered at the end (and that's NOT only because Spot is so cute!)
I enjoyed this class. Sometimes, though, I think we spent too much time on a single topic/overhead. The homework should be a little easier. But, the material was very interesting and I liked it overall.
Maybe simulations could be written for various techniques in order to demonstrate results more effectively.
The 1st 2 weeks were sluggish and uninterestng. The best part of the lecture was this last week - artificial life. You should have spent more time on that section and not so much with the others. As I mentioned before, it is very beneficial to know how this stuff is used. It helps to make the student aware of why they are learning something. These are just suggestions, and I hope they help. :-)
The assignments were very trivial problems in math and seemed pointless. The algorithm didn't seem very useful when we had to implement by hand. It might have been better for just concepts or [??] computer implementation.
At times, the packets skipped over information I felt was essential to completing the assignments. The time we spent on neural networks really helped me in understanding my team project.
Information covered was too shallow for any gain to have been realistically possible. Neural nets are interesting, but realistically, I learned practically nothing about them from this course. Work outside class was trivial, pointless, and irritating. I see no reason for us to manually perform a computer's tasks. Theory is fine, but there is a reason that no person does this by hand in the real world.
I would like to have acquired a better understanding of how neural networks function. In class and assignments, we primarily focused on perceptrons; the actual learning algorithms of neural networks are still unclear to me. The assignments seemed unnecessarily tedious, honestly. It may be better to explain the algorithm and let the class play with software that implements it (rather than using it themselves). Otherwise, I was impressed with this course as a first attempt.
I think you should cover less but more in-depth. I would have nejoyed more of that genetic algorithm evolution stuff and a complex study of ANN. I wouldn't change the assignments but no more last-week assignment.
I would have enjoyed the cours emore if we spent more time on neural networks and less on the other stuff. Also, more assignments with actual c++ implementations of machine learning techniques.
I wish we could have discussed more about artificial intelligence - not the ``Spot'' game, but about the more complicated ways that a machine learns. For example, the regression equations we did 1st week wasn't too interesting at all.
It would be very interesting if we could actually try these techniques/algorithms on a computer and obtain our own results for large data samples and find out how well these methods actually work.
I would have liked to know more about how researchers and computer scientists developed these techniques and the theory behind them. I thought the assignments were good.
Less time on data mining, more on artificial life and genetic algorithms.