Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Fri, 29 Jun 2012 16:35:45 +0200
From: Simon Marechal <simon@...quise.net>
To: john-dev@...ts.openwall.com
Subject: Re: markov per-position / masks

On 29/06/2012 13:32, Solar Designer wrote:
>> > +29% efficiency is great.
> Yes, I've seen a similar efficiency increase when I added per-position
> stats to John's incremental mode in 1997.  We have those since.

>From what I see the stats are trained on rockyou, and they are used to
crack rockyou. This means the the training and test sets are identical,
which means that there will always be an improvement (something about
the value of information that can't be negative).

It seems likely that character positions help, but the 29% figure is not
obviously informative.

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.