Ponezorg
March 7th, 2008, 02:59 PM
So what I want to do but am not sure how to do is create an algorithm for filtering certain data from a neural network I don't want all the data accumulated by the CPU to be written to the network just the neccessary information.....
This is of course bieng very vague....Ok I have a neural network....I created neural network class for the neural network object for data input 'X'....Sokay so one of the classes I want to link to a video input....which could use face and pattern recognition software to identify objects in the real world ultimately I want it to be able to learn but there must be a way to filter unnecessary information so it dosn't "think itself to death" so to speak....How would the source code for "Viedeo input" and filtering work.....
Lastly on a low priority note I would like the "sentient bieng" to be governed by isaac Asimov's three laws (just in case right, couldn't hurt) which could mean that the output is filtered using arithatic or by checking it's output to a hypothetical "law" neural network class...
$heap->{N.N.C} = $neural_n.n.c #neural network class
$heap->{N.N.O} = $neural_n.n.o #neural network object
$heap->{name} = $neuron_name;
$heap->{threshhold} = $threshhold;
$heap->{low_fire} = $low_fire;
$heap->{high_fire} = $high_fire;
$heap->{value} = 0;
I am currently programming in perl but am familliar with java and it seems easier when programming learning algorithms....I am familliar with lisp but it needs linux to run and I don't have a copy of knoppix to repartition my hardrive....
This is of course bieng very vague....Ok I have a neural network....I created neural network class for the neural network object for data input 'X'....Sokay so one of the classes I want to link to a video input....which could use face and pattern recognition software to identify objects in the real world ultimately I want it to be able to learn but there must be a way to filter unnecessary information so it dosn't "think itself to death" so to speak....How would the source code for "Viedeo input" and filtering work.....
Lastly on a low priority note I would like the "sentient bieng" to be governed by isaac Asimov's three laws (just in case right, couldn't hurt) which could mean that the output is filtered using arithatic or by checking it's output to a hypothetical "law" neural network class...
$heap->{N.N.C} = $neural_n.n.c #neural network class
$heap->{N.N.O} = $neural_n.n.o #neural network object
$heap->{name} = $neuron_name;
$heap->{threshhold} = $threshhold;
$heap->{low_fire} = $low_fire;
$heap->{high_fire} = $high_fire;
$heap->{value} = 0;
I am currently programming in perl but am familliar with java and it seems easier when programming learning algorithms....I am familliar with lisp but it needs linux to run and I don't have a copy of knoppix to repartition my hardrive....