Main Hint:

This hw is related to the equations discussed in lecture a LOT..
Most of the questions are straight forward to answer if you are
comfortable with them. So make sure you understand the perceptron
lecture..

1. Play around with the equations.. See if you can get a different
result if you make the suggested changes..

2. Same as what Bart did as an example in class.. Dont forget to add
   the threshold input and weight!

3. The domain of these functions is NOT binary! They are boolean values
   functions since they return YES/NO.. You can regard the domain as
   real values.. Think in terms of graphs.. The strictly greater property
   can be translated into something else which is easier to answer..

	This hint was wrong.. see solns and the newsgroup.. The domain
	is indeed binary and there were only 4 cases to test <0,0>
	<0,1> <1,0> <1,1>..

4. This question seems familiar.. The dec tree example in the previous hw
   was indeed larger than optimal.. BUT still try to come
   up with the general idea of why this might happen.. we DONT expect you
   to find a procedure to force the algorithm come up with arbitrarily large
   trees

5. Do what the hint says.. Time to explore what "linearly separable" really means..