I suspect that, statistically, there is no difference in the percent of victims harmed by criminals using military-style semiautomatic assault weapons during the ban, before it, and after it.
Initial indications seem to show that the number of officers shot with rifles went down significantly after the ban expired. Whether this is true of "military-style assault weapons" or not probably depends on how carefully you define military-style assault weapon to skew the numbers. There aren't many law enforcement deaths resulting from rifle shots at all, so it would be very easy to lie with the numbers by selectively narrowing the category further.
Also, doesn't the Brady Bunch count shootings in self defense as homicides?
Basically, yes, they call self-defense murder:
http://www.concealedcampus.org/bradycampaign/