There will be three homeworks and one mini-project, all of which will count towards 50% of the total grade. Since we would like to be able to make solutions available as soon as possible, we will be strict about lateness. The late policy is as follows:
| Days late (not including weekends) | Penalty |
| 1 day | 10% off |
| 2 days | 30% off |
| > 2 days | Not accepted |
The hand in time will be the beginning of class ie. 1:30 pm. The days described above are 24 hour periods starting from 1:30 pm of the due date. Note that since most homeworks are due on Thursday, the late penalty is 10% for hand in before 1:30 pm Friday and 30% for hand in before 1:30 pm on the following Monday.
Handed out January 25, due February 10
Download Homework 1 here.
The homework states: "A more efficient search method also checks for equivalences. In other words, a state may be reached via different paths from the initial state. For extra 5 points credit, ensure that only the shortest path remains alive in the search when two paths from the initial state merge."
I read that to mean checking for shortest path to a given state *as you go* ("remains alive in the search..."), instead of exhaustively searching and then printing out the shortest path.
Handed out February 10, due February 24
Download Homework 2 here.
There have still been many questions concerning questions 1B and 3, and an error in the assignment.
1A - For number 12, the english language version does not actually match the 1st order logic provided. Ignore it. For your edification, a correct 1st order logic interpretation of the English would be: All(x)[Treefrog(x) => Exists(y)[Car(y) ^ Drives(x,y)]]
1B - When trying to prove that a statement does not have a resolution, any complete proof will do. In addition, I think that some may have misinterpreted my statement that you should "try all possible resolutions" to mean that you should disregard whether resolving your current resolvant with a given clause will actually help in any way. It is sufficient to try only those resolutions which help you; i.e. those which cause the elimination of one of the predicates in the previous resolvant. However, you must show all of _these_ possible resolutions in order for it to be a correct proof. I apologize if this led any to do lots of busy work which wasn't necessary.
3A2 - In this question, think of what you're supposed to do whenever a statement is assigned the value you assigned it in 3A1.
3B - I have stated previously that ? means we do not know which of the other possible values a statement is taking. Note that in this case, there are 3 other possible values, T, F, and @, and ? can mean any of the three.
3B - Also, note that if a variable's value is @, it means that the variable cannot be true all of the time (otherwise, we would say that it has value T), and it cannot be false all of the time. That is, there must be _some_ fluctuation in its value.
Handed out March 14 due April 4
Optional early hand in March 24, at 4:30, NSH 4517 for 10% extra credit
Download Homework 3 here.
Download Neural Nets tutorial here.
Donwload the Addendum to Homework 3 here. (Last updated March 24.)
There was another small problem with one of the code files. If you've already downloaded the tar file, you can get a copy of just the updated file here:
/afs/andrew.cmu.edu/scs/cs/15-381/381/public/util-bayes.c
If you haven't already downloaded the code, the tar file has been updated with the new version of the file and you don't need to worry about this.
Handed out March 23, proposal due April 6, final write up due April 27
Download the Mini Project description here.
A practice midterm is available here
Solutions to the midterm are available here
Solutions to the final are available here