Copy Link
Add to Bookmark
Report
Xine - issue #3 - Phile 114
/-----------------------------\
| Xine - issue #3 - Phile 114 |
\-----------------------------/
The Schumrz algorithm of genetical mutation
And its possible applications
Once again by [Star0]a.k.a.[Unkm/Ikx]
I was two weeks ago at the university with my class , there was a lot of
stand I look from there to there , when I saw the Shurmz stand, there was
a sillicon grafx running with the shurmz game
What are the shurmz ? Shurmz are pretty little ants who search food ,like
Sim ant , the new stuff is that they can mute their specificity , there's
six or seven specificity like speed , force , intelligency , resistance ,
reproduction etc etc...
The game is simple,you have two colony ,black and white,if an ant doesn't
find enough food,she dies,an ant can die also when two ants fight and she
lose the round.There begins social-genetical mutation , if the colony one
begin to have more force,so only ant who have speed and force 'll survive
in the colonny 2 , and the reaction of the colony two is to produce more
child with a good force and a pretty speed . Nice no ?
The most interesting things in their stand was the shurmz algorithm of
genetical mutation when reproduction came , in that case you have 3
important steps :
Ú- Reproduction ¿
| Ú-------Á-------¿ 1ø Selection , establish the best
| | Selection | person who can respond perfectly
| À-------Â-------Ù at a present problem, in some word
| Ú-------Á-------¿ the most adapted at the situation,
| | Crossbreeding | this person'll have more children
| À-------Â-------Ù than other
| Ú-------Á-------¿
| | Mutation | 2ø Crossbreeding , this step
| À-------Â-------Ù consists to pick the best of two
À---------------Ù persons like sexual reproduction ,
generating an individual person
3ø Mutation , may be the most complex thing in their algorithm , the
mutation scheme is at my vision chaotic , randomic . But there can be
mathematical , dunno how , but this may more possible . So the mutation
generate some gene randomic , in that case there's two possibility , die
or live , when the life live , there's new gene that can help / handicap
if the person is handicapped by this gene ,so this gene is skiped in the
crossbreeding , can be ignored at the step 1 and it dies completely, gene
lost . But in case of help , the gene will be spread over a maximum of
people over generations
Interrest for VX ?
Sure , mutation is a sweet dream for all vx writer , the idea that a Com
infector mute himself into a winexe infector is really pretty but kinda
impossible,mutation are too complex to be applying on a virus,and anyway
the number of mutation is so high to be capable to see if anything where
changed, and if we increase the number of genes who mute each time,we'll
have too many handicaped children and the colony goes to death , but
there's a pretty application of this scheme into virus comportement ,like
how many infection per process , type of infection ( if per process or
goes resident ) which platform ( if there are many ) ,this gene can mute
a lot , look , if any stupid aver want test yar virus with a lot of bait
file,the infection rate'll descend a lot and goes to 1/8 , why ? because
the virus adapt himself on external modifications , analysis become more
and more difficult because viruses are placed on unusal environments ,
and in this case the virus give reaction to rest in a kinda normal state
,but when the virus'll return on a normal machine , his reaction returns
normal ,normal from the enviroment now imagine that the virus encounter
a new kind of his virus branch
The virus from left is a lame comexe and the virus from right is a good
exesys what's happend? a new virus born,it's an comexesys who he'll kill
all his fathers and mothers and take their place , there's the bad point
of computer virus , they don't know the natural death , it's a primordial
point in the biological evolution ,only perfect virus can never die ,but
you know that nobody is perfect . Return on the virus crossing problems .
The first problem is if the comexe are really too bad and 'll handicap
the comexesys . It crash over the virus,or is detected by an antivirus ,
think that the comexesys have already killed all comexe and sysexe . So
the colonny have suicided itself , it's stupid , and the nature give a
good reponse against this problem : dissimination . The comexesys don't
replace all his parents by himself ,but 1/3 father,1/3 mother 1/3 child ,
and if the comexesys goes right , his parents population goes down , the
best adapted'll survive. The second problem is how to detect that the exe
infection from good exesys is better than the lame comexe , the best idea
is to define a protocol for viruses, and define infection level,low equal
lamah , high equal good , in this case , viruses can choose with which
routines they can use, too bad level can be prohibed, any other solution
of this problem is welcome . The third and last problem is the
imcompatibility between some kind of infector,a com and a win32 infector
can't couple themself, in solution of this problem,you can create bridge
of compatibility from system1 to system2 .
An important rule is that the family must respect a minimum of rules ,
like active detection , or let's a better virus install himself at his
place, already infected detection , etc etc otherwise that crash.The best
idea I think is to apply the concept on polymorphic engine , they'll
always modify themself , so when an antivirus have defeated the
polyengine1 , he have already muted himself in anything different , and
life continue . In this case , viruses becomes allways better and better,
and always couple old tech and new technology , there also appears
populations of kind of viruses,in this case appaers also complex movement
of infections , see automata
Les petits dlinquants (C) Star0'98
