|
Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com |
From: Bawcom, Aaron (aaron_bawcom
intrusion.com)Date: Wed Jun 13 2001 - 22:59:46 CDT
Archive: http://msgs.securepoint.com/ids
FAQ IDS: http://www.sans.org/newlook/resources/IDFAQ/ID_FAQ.htm
FAQ NIDS: http://www.ticm.com/kb/faq/idsfaq.html
IDS: http://www-rnks.informatik.tu-cottbus.de/~sobirey/ids.html
HELP: Having problems... email questions to ids-owner
uow.edu.au
NOTE: Remove this section from reply msgs otherwise the msg will bounce.
SPAM: DO NOT send unsolicted mail to this list.
UNSUBSCRIBE: email "unsubscribe ids" to majordomo
uow.edu.au
-----------------------------------------------------------------------------
When an AIM-9J Sidewinder leaves an F-15, the plane does not decide to
launch the missile, the fighter pilot does. Information systems will
continue to augment human choices. Like when Sigourney Weaver uses that big
mechanical heavy lift body machine to fight the Queen in Aliens 2, so will
information security tools enhance & empower the decision making
capabilities of the human mind. Algorithms will make more and more "leaf"
decisions which will continue to build on each other. There has been some
fantastic research on the subject of tiered statistical analysis systems
where different algorithms build on each others strengths. For example:
immediate chaotic data is grouped into peaks using fuzzy logic. The peaks
are then parameterized into a neural network to distinguish categories. The
categorical information is then applied to case based reasoning models which
then identify extremely high level patterns of behavior based on empirical
example. All of these smaller decisions are controlled by highly informed
master choices that are exposed to the user in the most comfortable way
possible.
"If an army of graduate students can do it, then Microsoft can do it"
-----Original Message-----
From: Kohlenberg, Toby [mailto:toby.kohlenberg
intel.com]
Sent: Wednesday, June 13, 2001 9:25 AM
To: 'Marcus J. Ranum'; Kohlenberg, Toby; 'Steve Robinson';
Peter.Watson
sunlife.com; ids
uow.edu.au
Subject: RE: IDS: RE: RE: IDS Future
Archive: http://msgs.securepoint.com/ids
FAQ IDS: http://www.sans.org/newlook/resources/IDFAQ/ID_FAQ.htm
FAQ NIDS: http://www.ticm.com/kb/faq/idsfaq.html
IDS: http://www-rnks.informatik.tu-cottbus.de/~sobirey/ids.html
HELP: Having problems... email questions to ids-owner
uow.edu.au
NOTE: Remove this section from reply msgs otherwise the msg will bounce.
SPAM: DO NOT send unsolicted mail to this list.
UNSUBSCRIBE: email "unsubscribe ids" to majordomo
uow.edu.au
----------------------------------------------------------------------------
-
Good explanation, thank you!
Is it fair to say you don't think it is possible to duplicate the
anomaly detection capabilities of the analyst, and therefore the
improvements must be in making the information more available to
the analyst?
I can certainly see that, it is the approach I see as having the most
value today, definately. I think that puts a different twist on the
question that was asked-
What do people think is needed/lacking in the analyst console?
And, a second question-
Do people believe that further development will make is more feasible
to have an IDS be effectively monitored by someone who is not a
skilled analyst?
Toby
> -----Original Message-----
> From: Marcus J. Ranum [mailto:mjr
nfr.com]
> Sent: Tuesday, June 12, 2001 11:09 PM
> To: Kohlenberg, Toby; 'Steve Robinson'; Peter.Watson
sunlife.com;
> ids
uow.edu.au
> Subject: Re: IDS: RE: RE: IDS Future
>
>
> Kohlenberg, Toby wrote:
> >would like to see some
> >serious attempts at "artificial intelligence"/neural net-type
> >monitoring of events
>
> There _have_ been serious attempts. Lots of serious attempts!
> Just because they haven't exactly worked, doesn't mean that
> they weren't serious. ;)
>
> Conceptually, there are some problems with the whole concept
> of using neural net-type applications to do broad-based anomaly
> detection. You can use them to do pattern detection in the small,
> where you build small baselines relevant to a particular knowledge
> bases. But in that case, what you're really building is an expert
> system with statistical analysis at the leaves of your decision
> tree - semi cool but not "artificial intelligence."
> mjr.
> ---
> Marcus J. Ranum Chief Technology Officer, NFR Security Inc.
> Work: http://www.nfr.com
> Play: http://www.ranum.com
>
>
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]