System Performance Degrade over time
- Thread starter Meatwad1337
- Start date
- Status
RicoDirenzo
Fully Optimized
- Messages
- 2,107
Stimulus/response is an example sting logic. String logic is that all actions are inherenetly interconnected (strung together) with each other...like the brain.....programming is a string of interrelated commands all connected to goal or funciton of the program. If you remember DOS commands and the error: divide stack overflow.....DOS reads an instruction outside of the string of internal commands and program logic and whammmoooooo.....Sorry, Will Rodgers, that does not compute!!! (lost in space robot Robbie) In other words, the command lies outside of the internal logic of the program and the OS goes to brain death...Blue screen of death is an example. :0 or too much time with the kegerator!!! 
Whew, I took an entire class on just this stuff once, so bear with me, I'll try to distill it:
Strings are just... strings. Call it "input" to your program...
Let me define a string "aabbbbbbababab"
It doesn't mean much by itself, its just a string.
A "language" is a pattern (the language is like what your program is looking for) that can be understood by a "machine". So, a turing machine is just a theoretical construct of a "machine" that involves an infinitely long tape that stretches in both directions. The machine has a definition (a string that defines how it behaves) and it reads the tape. So, read head moves right, reads a value from the tape, depending on the value, moves some number of steps left or right, and reads again. The definition of the machine (like the source code of the program) defines the behavior of a machine.
A machine is designed to be able to "recognize" a "language". There are certain classes of languages that are "machine recognizable". This really goes beyond the scope of the discussion, but suffice it to say, any program that can be computed by a computer is "turing recognizable". (There are also lesser forms of machines that can interpret simpler languages. The simpler languages might be called "regular expressions", etc)
Anyway, so, if I define a machine by a string (trust me, its possible, but the way this is done is beyond the scope of this discussion), I want it to parse the input string, evaluate it, and IF it "belongs to the language defined by the machine", then I halt and "accept". If it does not fit my language, I halt and "reject." (Pass/Fail)
So, the language:
ab*a*b
This language just defines a language that begins with an 'a', has 0 or more b's (* is the Kleene star, which denotes that a character may occur 0 or more times), has 0 or more a's, and terminates with a b.
So if I define my machine to "accept" that language, then if I input:
aaabab
The machine rejects, because it does not meet the language. (Too many a's at the start)
If I input:
abbbaaab
The machine accepts, because it fits the language.
And, believe it or not, you can build ALL of modern programming theory out of these simple ideas. Its staggeringly complex though. I hope that made some sense.
Strings are just... strings. Call it "input" to your program...
Let me define a string "aabbbbbbababab"
It doesn't mean much by itself, its just a string.
A "language" is a pattern (the language is like what your program is looking for) that can be understood by a "machine". So, a turing machine is just a theoretical construct of a "machine" that involves an infinitely long tape that stretches in both directions. The machine has a definition (a string that defines how it behaves) and it reads the tape. So, read head moves right, reads a value from the tape, depending on the value, moves some number of steps left or right, and reads again. The definition of the machine (like the source code of the program) defines the behavior of a machine.
A machine is designed to be able to "recognize" a "language". There are certain classes of languages that are "machine recognizable". This really goes beyond the scope of the discussion, but suffice it to say, any program that can be computed by a computer is "turing recognizable". (There are also lesser forms of machines that can interpret simpler languages. The simpler languages might be called "regular expressions", etc)
Anyway, so, if I define a machine by a string (trust me, its possible, but the way this is done is beyond the scope of this discussion), I want it to parse the input string, evaluate it, and IF it "belongs to the language defined by the machine", then I halt and "accept". If it does not fit my language, I halt and "reject." (Pass/Fail)
So, the language:
ab*a*b
This language just defines a language that begins with an 'a', has 0 or more b's (* is the Kleene star, which denotes that a character may occur 0 or more times), has 0 or more a's, and terminates with a b.
So if I define my machine to "accept" that language, then if I input:
aaabab
The machine rejects, because it does not meet the language. (Too many a's at the start)
If I input:
abbbaaab
The machine accepts, because it fits the language.
And, believe it or not, you can build ALL of modern programming theory out of these simple ideas. Its staggeringly complex though. I hope that made some sense.
TheHeadFL said:
I could of told you that! Oh man, thats heavy. i'm going to have to reread that a few times before I know completely what you mean. I got the basic idea of a turing machine though. Based on one set of instructions, it goes to the enxt set which tells it where to go and so on.
I'm digging the knowledge, feel free to throw as much my way as you'd like. You dont have to dumb it down either, leave the deciphering of it up to me.
RicoDirenzo
Fully Optimized
- Messages
- 2,107
Humans, being conscoius of themselves and outside stuff, have a very sophisticated string logic board. It's called the central cortex...i.e., both sides of your cerebrum........now for the mind blowing part. Intel CPU's contain (I really don't know) but lets say 5x10 to the sixth transitors/interconnections. 5 million. The average human brain has 18 billion nerurons, and their average interfaces contain an addtional potential 10 to the sixth swithes (on and off swtiches that interface with other neurons).
so...18 billion x 1,000,000. Lets build a CPU with current micro co-prossesor (SP?) technology. One MIT electrical engineer made this statement given the above information about the human brain: In order to build a CPU with that many interconnections it would be 10 stories tall and cover the entire state of Texas. Goose bumps anyone???................and some moron's actually believe this was accomplished by natural selection. ROLF!!!!!!!!! Try and write some code to that system!!!!!
so...18 billion x 1,000,000. Lets build a CPU with current micro co-prossesor (SP?) technology. One MIT electrical engineer made this statement given the above information about the human brain: In order to build a CPU with that many interconnections it would be 10 stories tall and cover the entire state of Texas. Goose bumps anyone???................and some moron's actually believe this was accomplished by natural selection. ROLF!!!!!!!!! Try and write some code to that system!!!!!
Its not that unbelievable really, when you consider that computer scientists are currently borrowing heavily from neuroscience in their research. The so called "neural network" machines already are capable of finding patterns and doing other such non-linear type work with unparalleled speed when compared to linear-based computational algorithms.
In a laymans opinion (since I am admittedly not formally trained on neuroscience in any way) there is no reason that that functionaility has to be contained within one system. Many millions of computers are already in existence, and one doesn't neccesarily need to approach the level of complexity of the human brain to approach it in terms of achievable useful output. (Since we all know how human beings aren't working at capacity)
Like individual neurons, each linear-based computer is dumb, in and of itself. However when combined with millions of others in a meaningful way, useful behavior can be observed. Does one ant really know how to build an ant hill? Of course not, but a million ants collectively exhibit something of a collective consciousness. Building a computer with the non-linear cognitive ability of even an ant colony would be an achievement, considering their skill at architecture.
Of course, the real philosophical question is, can the brain even be modeled mathematically, however complex it may be? While the math may be out of reach even if we could, I don't think people are ready to acknowledge that the brain could be something like a finite state machine. That is, people do not like to believe that given the input, and the state information, the output can always be exactly predicted.
Given a complete knowledge of brain chemistry, as well as a complete "snapshot" of all the neurons and their interconnects in a human brain, combined with the input to the system, in my opinion, probably doesn't "define" the human brain. The consequence of all of this would mean that if it could be approximated by a finite state machine, than human behavior is totally predictable and nonrandom. I have trouble with this notion.
I do believe that computers will continue to make better use of neural network computing however, given its unparalleled pattern recognition abilities.
In a laymans opinion (since I am admittedly not formally trained on neuroscience in any way) there is no reason that that functionaility has to be contained within one system. Many millions of computers are already in existence, and one doesn't neccesarily need to approach the level of complexity of the human brain to approach it in terms of achievable useful output. (Since we all know how human beings aren't working at capacity)
Like individual neurons, each linear-based computer is dumb, in and of itself. However when combined with millions of others in a meaningful way, useful behavior can be observed. Does one ant really know how to build an ant hill? Of course not, but a million ants collectively exhibit something of a collective consciousness. Building a computer with the non-linear cognitive ability of even an ant colony would be an achievement, considering their skill at architecture.
Of course, the real philosophical question is, can the brain even be modeled mathematically, however complex it may be? While the math may be out of reach even if we could, I don't think people are ready to acknowledge that the brain could be something like a finite state machine. That is, people do not like to believe that given the input, and the state information, the output can always be exactly predicted.
Given a complete knowledge of brain chemistry, as well as a complete "snapshot" of all the neurons and their interconnects in a human brain, combined with the input to the system, in my opinion, probably doesn't "define" the human brain. The consequence of all of this would mean that if it could be approximated by a finite state machine, than human behavior is totally predictable and nonrandom. I have trouble with this notion.
I do believe that computers will continue to make better use of neural network computing however, given its unparalleled pattern recognition abilities.
I should note, however, that this is as long as we continue to use deterministic computers. That is, computers that can't execute all code paths in parallel.
A nondeterministic computer could execute all possible code paths at the same time. This would open up all sorts of possibilities with neural networks and would probably mean that a neural network that parallels the human brain in computational ability (though not neccesarily consciousness) might even fit inside a building, let alone a room.
Here's hoping that some day in my lifetime there will be a quantum computer....
A nondeterministic computer could execute all possible code paths at the same time. This would open up all sorts of possibilities with neural networks and would probably mean that a neural network that parallels the human brain in computational ability (though not neccesarily consciousness) might even fit inside a building, let alone a room.
Here's hoping that some day in my lifetime there will be a quantum computer....
RicoDirenzo
Fully Optimized
- Messages
- 2,107
Non-deterministic computers...hehehehehe.....an academic way of saying: "Computers with free will". Wow!!!! Here come the Cylons.........guess we need to start building the Battle Star Galactica!!! Don't think that will happen too soon!!!!
No, non-deterministic computers don't have anything to do with "free will". You're misreading my description.
They are computers that can compute all possible solutions to a problem simultaneously, rather than one branch at a time.
Determinism just refers to whether or not execution can be expressed in linear terms. Nondeterminism implies parallelism. Nothing too unbelievable about that.
They are computers that can compute all possible solutions to a problem simultaneously, rather than one branch at a time.
Determinism just refers to whether or not execution can be expressed in linear terms. Nondeterminism implies parallelism. Nothing too unbelievable about that.
RicoDirenzo
Fully Optimized
- Messages
- 2,107
I understand the difference between multi-tasking systems (mainframes already do this) and what you call non-deterministic/parallelism. Your gonna have to trust me on this one. Having taught college philosophy our discussion has now entered into a Philosophical discourse called epistomology: The definition of and understanding of knowledge/consciousness. non-determinism in philosophy has to do with self awareness and free will. The thing/being in itself chosing a specific set of options or option regardless of the pre-existing information, motive or understanding. That's a huge thing!!!!! It also, at this point, crosses into the gray area of theology......the issue of pure materialistic reductinism vs the creative engineering implied in the "first cause". Far to complicated, obviously, for a tech forum. But you are pretty impressive thinker Dude. Hats off to you. On my way home. It's been a great day on TF!!!!!
- Status