Wednesday, October 12, 2011

Problem solving

Introduction:  Couldn't find an image of fog on a street?  Arad to Bucharest?  This example is straight out of the text.  It would be better to give a different example, so the student can see two examples instead of the same one twice.

What is a problem:  text getting a little cramped there.  How is this better than a power point slide?

Tree Search:  There are numbers on the edges, we know they are distances, so why are the lengths just # of hops?  Needs more discussion, especially with breadth first search-  You say it's shortest first, but in the algorithm you ignore the path cost- why?.  Wait, did we just re-write the tree search algorithm on the page to make it more cramped?  "The reason we keep track of the explored state..."  we just covered that and you argued that in tree search it is a natural way of looking at the problem to allow duplicates, but now you're saying we prune them out?  Which is it?

Graph Search: This whole transition from tree to graph was really awkward. 

Say wouldn't it be great if I could see the code as you walk through the algorithm? Also would be helpful when I'm trying to answer the quiz questions.  Still haven't discussed the difference between path length and path cost.

Is unit 11 broken?  I can only play half of it.

Is he playing with candy wrappers? 

Uniform cost: woah he just totally missed the discussion of whether to add Oradea from Sibiu. That's important.  Omitting it is a flat out error.

Search Comparison:  He said I was supposed to fill in the numbers of node expansion order, but I never got a chance to do so.  Wait, we're looking at the node expansion order of DFS w/o first looking at what the algorithm is?

More on Uniform Cost:  I like this.  The topological map model of how the search expands was a nice explanation for greedy best first search.

A*: no proof of optimality, just a hand wave?  Ooops, I made an arithmetic error on the quiz :-(

Peter needs a better 15 puzzle toy.  Those tiles don't move very well.

So, how would you relax the problem in the path finding problem?  how could the program get those numbers?  The argument that we can automatically generate heuristics only works in the sliding tile puzzle.

6 comments:

  1. It feels like you are being overly critical. What's your goal here? Since it's certainly not constructive.

    "Peter needs a better 15 puzzle toy"

    I mean, seriously, why don't you just insult his haircut and call his mother a whore and be done with it?

    ReplyDelete
  2. I agree ... give them a chance. Yes, the questions are ambiguous, but perhaps they'll listen and clarify more in future, perhaps with some kind of formal spec for each question. Your blog is good, and voices the frustrations many people have ... cheers.

    ReplyDelete
  3. It seems to me the whole point of this blog is to be harshly critical. Obviously staying as professional / constructive as possible is the ideal, but there is no necessity for it. We all know and accept that we haven't paid anything to have access to this class and that we are guinea pigs for a new thing, but time is money and we expect a certain standard from Stanford and the highly acclaimed professors. Speaking out, sometimes vehemently is the best way to promote improvement.

    I'm glad I'm not the only one who was wondering about the candy wrapper sound.

    As someone who has already taken this course with this text, I've found the explanations, teaching, and especially in video quizzing pretty weak. But I'm hopeful for improvements as they get a hang of things.

    ReplyDelete
  4. Yes, it's true, this is free, but our time has a cost too, we are investing our time (and hence money) in this course because it was promoted by a prestigious university like Stanford.

    And please, we shouldn't be so ingenuous, this course is not free just for charity, do you really think Stanford is not befited from this? is a huge promotion for them, even the teachers are selling thousands of books thanks to this course.
    And it's nothing wrong with that, but at least please raise the level !

    Even more, this could be very counter productive for them if they continue in this way, I wouldn't pay a lot of money for studing in Stanford with classes like these.
    In my country the education is public and free, and even here we have a higher level.

    Anyway, I agree that you (the author of this blog) should put your corrections, don't make it only critical.

    ReplyDelete
  5. Hi I'm doing this curse to learn a lot of things and like you I've got some points I'm not comfortable with but in general I think it's great to get knowledsge for deeper investigation. But probably you can produce such a course too. Or post a recording of your lessons on youtube. Probably your students and of course the whole world will profit from it. If you're doing so, please give me a mail, I'm very interested.

    sunny_simon@mail.ru

    Thanks a lot for your comments and keep blogging, but a little less critical and a bit more constructive.

    ReplyDelete
  6. The point of this blog is to truly live-blog the videos. I just write what I'm thinking as I watch. My comments have been overwhelmingly negative because I have been greatly disappointed. I am disinclined to cut them much slack for the following reasons:

    1) These are very smart, very experienced people with world-wide reputations, one of whom (Peter Norvig) has demonstrated that he can generate careful and clear instruction in several books.

    2) They and Stanford played the PR of this out as much as they possibly could. By doing so, they asked for careful scrutiny.

    3) This really is an opportunity to try a novel educational model that only a few have the resources for. They have an obligation to do the best job possible.

    I will however endeavor to be more constructive as well.

    ReplyDelete