Brain Neural Partitioning
100,000 neural packages is a lot to handle. To make it more manageable, two partitions of approximately 50,000 will be created. This will keep neural injected signals clean, make programming more easy, and maintain a commonality of interface.
These photos show the complexity of neural net wiring. A variety of color coded wires were used as the initial alotments of 1,000 wire cables were expended.
It takes only two Propeller chips to inject 100,000 neural machine packages. This is a very efficient design. Expansion is possible with a Propeller chip controlling each additional 50,000 machine neurons. Expandors are in increments of 50 Propeller chips. Expansion is unlimited.
The Giant Brain has now increased neural net partitioning from two 50,000 units to three. The third partition is currently open ended and undergoing fulfillment and testing. The first two partitions are completed and undergoing tests and configuring.
Brain Breakthrough! 100,000 Neurons Exceeded Moving the Giant Brain to the next generation
Exceeding 100 Propeller chips and adding the third partition breaks the 100,000 barrier. There is really no limit now. Expect a new generation of high density neural nets.
No Photo's in post #543.And a belated congratulations to Your 100,000th Neuron baby...
Ttailspin, thanks! I just took photos today. They are waiting for my computer to come back from the service center so there is a slight delay in posting. I'm making most of these posts with an iPhone. I'm trying to decide which new computer to buy. Maybe Apple dual boot with OSX and Windows.BTW, the photos look fantastic but I'm a bit burned out from all the wiring.
What type of ANN is it? What problems is it suitable for?
Leon, thus far I've put up a layer ANN and this week began the partitioning to make over 100,000 neural delivery packages more manageable. The neural net injectors needed a fair amount of work and those are smoothly working this week. The wiring will gety some cleanup and a more permanent fixture for the breaadboards is needed. It's an enjoyable hobby project that's taking me somewhere and I cannot be certain about the final result. I built it for fun and seeing what it can do. It may not solve any problems. I still think the primary objective is to bring some life into the brain, though the exact type and nature of this machine life remains undecided. I also want to try some new things with it, like enhancements, new algorithms, new interfacing, new communications, and forthcoming ideas. I originally began building this brain for an advanced humanoid robot so that's the end result. For now, I want to improve it, test drive it, and see where it can take us..
The question was asked about the type of ANN the Brain is using. The Brain is running a sample neuron package called Neural Material or Neural Package. This is really a unique demonstration of a Neuron
Primative. What is it? What can it do? What purpose does it serve? The purpose of this demonstration Neuron is to infiltrate the Brain to a maximum density using the Neural Injectors. The next purpose is to fire each Neuron with a periocity of specificity. The next purpose is to exemplify that neural "reaction." Undoubtedly additional parameters are being explored in these early primatives.
Brain downtime is incurred as the engineering computer containing all the Brain software has gone bad. It was apparently hit with a virus. Semantic Anti-Virus was disabled and would only scan and report virus. Another Virus program was installed and it deleted 27 virus and quarantined 56. Unfortumately, Windows explorer.exe and EXPLORER.EXE was deleted. Upon reinstall, the desktop returned but the OS would never work after that. More virus was identified that the antivirus program did not delete including startup.exe which was a hidden file Trojan-Worm. Several computer techs have worked on this for three days, are working on it today and last I heard the entire Windows OS was corrupt and will need to be installed. A reinstall is scheduled for today. It is currently unknown if the hard drive and nearly half a million files have survived the ordeal. Let's hope for the best. As the computer is five years old, it's time to replace it. Any ideas for replacing it with Mac which may be a little more virus resistant? Or maybe too many programs need PC. I am also thinking about a dual OS Mac with OSX and Windows. Any advice is appreciated.
Considering the importance of the Brain project and the immensely invaluable functions that a supporting computer must provide, and the general inability to tolerate down time, a new consideration is in effect for providing a new dedicated Brain support computer.
The dedicated Brain Support Computer will have the following requirements:
solid state
windows and linux os capable
bootable from usb stick drives
internet capable
massive external usb hard drives with removeable storage
The actual Brain signal divide points where the number of Propeller chips begin to see a degradation in signal quality during parallel loading is at 56 props. This divide point is noticed at loading when the number of 56 is greatly exceeded. Additional signal conditioning may easily remedy this, though a partitioning scheme is currently chosen to immediately assure high signal quality. Tests will be run at a later date with more signal conditioning. Tests show that once the code is loaded into multiple partitions, props are free to exchange data on a net. 17x3+2+3=56 Since each partition is "driven," the number of partitions is unlimited or based on whatever is manageable. Remember that current configurations of three partitions up to 50 props each will deliver 150,000 packets of neural material.
To meet all requirements for a dedicated portable Brain Support Computer DBSC, a loaded EEE PC with the ATOM processor is chosen. Tests are now in effect to determine the efficability of massive external TB USB drives for program storage, and to pace the computer and perform setup of software. Brain code is nontransferable until the original supporting pc is repaired and returned.
It is noted in tests that the signal condition as determined by the Brain Signal Divide Point BSDP, is affected and regulated by the broadband signal of the breadboard. Some breadboards sustain and carry a more clear signal. This assessment is made with a variety of solderless breadboards tested from the following sources:
usa
china
taiwan
More tests are needed to ascertain the potential to increase the BSDP in the future.
A subjective analysis of the above mentioned solderless breadboards is as follows:
usa - 80% of signal
taiwan - 90 % of signal
china - 60% of signal
Signal testing is setup with arrays of Propellers to determine go/no go situations. When the original no-go situation was encounted, the number of props was backed off for the establishment of a reliable partition. The partition was built with a combination of breadboard sources.
Ubrupt Unfortunate Brain Project Termination
Bad news - the Chinese MIS computer man just reported in with the status of my computer. He says somehow, when he was working on it, all files were deleted off the hard drive when "repairing Windows." He regrets this unfortunate mishap. This includes the deletion of the last 5 years of my work, including all Propeller Brain files! Hard drive inspection indicates it is cleaned. He removed the hard drive and it will be taken to another Chinese computer company tomorrow morning, apparently to see if they can find any files. I'm not optimistic about the outcome of this situation and will immediately give up all computers with Windows operating systems.
Humanoido, sorry to hear about your data loss. I've found uploading copies of source code to the forum is a free way to backup your code. I do that and upload my source code to Gmail as an attachment. Something to consider for the future.
My ISP has a "Digital Vault" for backups; I get 5 GB free (50 GB for a small monthly payment), which is plenty for me. I compress all my important files and upload them occasionally. I really must get round to writing a script to do it automatically on a regular basis.
Your data files should be fine. The only way they deleted is if someone ran FDISK or Format. Just have your service upload the files to an online service or drop them on an external HDD. Check out DropBox. They give you 2G free space and it's extremely convenient. There's also Google code.
What about the system you used post the last couple of threads? You can limp along with that until you get a new box.
Even with my assiduously backed-up backups, when the OS won't boot, the first thing I reach for is my Knoppix CD, boot up the the PC from that, and copy all my work files onto another drive. Only then do I consider the next step: reload the OS, or start over with a fresh drive. It's still much simpler, assuming the disk isn't completely toast, than restoring from backups.
Humanoido, I am sorry for your loss. I would very much like to have seen your "Neural Material" code. What unfortunate timing.
Your data files should be fine. The only way they deleted is if someone ran FDISK or Format. Just have your service upload the files to an online service or drop them on an external HDD. Check out DropBox. They give you 2G free space and it's extremely convenient. There's also Google code.
What about the system you used post the last couple of threads? You can limp along with that until you get a new box.
I posted the previous recent posts with an Apple iPhone. I'm now running eee pc for a short time.
Unless the virus deleted files... this morning he is saying the file names do not appear and wants to know the exact names of the files on the hard drive. My response is immediately return my computer to me. It was one thing to give permission to extract a specific Windows virus but another thing, not approved, to reinstall Windows and mess with the hard drive. The drive had 50GB capacity with (9GB free) approximately half a million files. Internet here is very slow, and most sites are blocked so any kind of online storage is not practical.
Even with my assiduously backed-up backups, when the OS won't boot, the first thing I reach for is my Knoppix CD, boot up the the PC from that, and copy all my work files onto another drive. Only then do I consider the next step: reload the OS, or start over with a fresh drive. It's still much simpler, assuming the disk isn't completely toast, than restoring from backups. Humanoido, I am sorry for your loss. I would very much like to have seen your "Neural Material" code. What unfortunate timing.-Phil
I also use a live boot LINUX CD, it's a bit older than the Knoppix but does the same thing. Apparently something happened to all the Windows files and data - this is what the service man is reporting today. My guess is he reinstalled Windows.
I keep all my development data and files on the desktop. I am told now that the hard drive shows 35gb on it but 11MB is missing and there are no file names for any data. For example, thousands of .doc files don't show up.
Latest update: I'm uable to get my computer back today. The language translator said the MIS man removed the hard drive and took it home and is working on it. The company CEO is now involved and put enormous pressure on the MIS man.
The company wants to give the man time today as an opportunity to work on it. There will be an update reported today in two hours. I think tomorrow is going to be a day of reconing.
I have left messages to have the hard drive put back into the computer and return the computer to me immediately. I will work on it myself but it may take 2 days to have the computer delivered.
Correct. So the real question is if a new Windows is installed over the old Windows, and the desktop now appears blank due to writing new source folders, where before it held all the folders of propeller files and data, and if the computer is booted with the LINUX CD, how to find and retrieve the hidden files?
Bummer, the Big Brain got sick and died... life is harsh.
At least You have much of Your notes about the Big Brain project right here on this forum...:cool:
Good points Ttailspin, As machines become more and more complex, the potential is born for more and more failures. Sometimes a machine becomes so complex, it defies troubleshooting. Many people have described Window'ed computers this way, not understanding the erratic and errant, often times intermittent strange behaviors.
One example, other than the Brain, is a complex humanoid that I was working on. I had the surprise of my life when I brought it to life for an extended period of time (charging and recharging batteries), then let the battery run down - big mistake! The humanoid flung out its right arm and hand then brought it back to the chest, as it it was having a heart attack.. it stumbled forward a few steps then fell limp to the floor in a twitching leg and full body spasm that was extremely disconcerting and worrisome to see. After that happened, I took measures NOT to let any humanoids die. Even in the BASIC Stamp SEED Supercomputer I wrote code so the ten life forms would not die. Unfortunately the Brain's supporting computer bit the dust before I could replace it.
Good point, I'm happy with the number of posts and the volumes of information made available on the Forum about the Brain and after I get my new computer I can address the possibility of reconstructing the software over time.
My only experience with The Knoppix CD has been to restore files from an (at least partially) intact file system. If your file system has been completely rebuilt, the Knoppix CD may not be of much use.
___________________
So let me get this straight. You've signed an NDA with a "publisher" who provided Spin code for some kind of neural simulation. Am I right so far? Or did they provide the algorithm, and you converted it to Spin? And, just to clarify, this "publisher" won't even let you say who they are? Do I have that part right as well? Also, what changes in a "couple months" that frees you from this NDA to publish raw code?
Humanoido, My only experience with The Knoppix CD has been to restore files from an (at least partially) intact file system. If your file system has been completely rebuilt, the Knoppix CD may not be of much use.
That's unfortunate about the files. I hoped booting a LINUX disk could be helpful. There must be some way to bring back these files since they are hidden but may still be on the hard drive.
I'm not asking you to discuss it, just to verify the circumstances peripheral to the alleged NDA. Humanoido, you've made some extravagant claims in this thread without much hard data to substantiate them. Please forgive my skepticism, but your evasiveness doesn't exactly bolster your credibility. A few crumbs of something substantial to chew would go a long way towards feeding our appetites and keeping us interested.
Comments
100,000 neural packages is a lot to handle. To make it more manageable, two partitions of approximately 50,000 will be created. This will keep neural injected signals clean, make programming more easy, and maintain a commonality of interface.
These photos show the complexity of neural net wiring. A variety of color coded wires were used as the initial alotments of 1,000 wire cables were expended.
[Add photos here]
It takes only two Propeller chips to inject 100,000 neural machine packages. This is a very efficient design. Expansion is possible with a Propeller chip controlling each additional 50,000 machine neurons. Expandors are in increments of 50 Propeller chips. Expansion is unlimited.
The Giant Brain has now increased neural net partitioning from two 50,000 units to three. The third partition is currently open ended and undergoing fulfillment and testing. The first two partitions are completed and undergoing tests and configuring.
Moving the Giant Brain to the next generation
Exceeding 100 Propeller chips and adding the third partition breaks the 100,000 barrier. There is really no limit now. Expect a new generation of high density neural nets.
And a belated congratulations to Your 100,000th Neuron baby...
Leon, thus far I've put up a layer ANN and this week began the partitioning to make over 100,000 neural delivery packages more manageable. The neural net injectors needed a fair amount of work and those are smoothly working this week. The wiring will gety some cleanup and a more permanent fixture for the breaadboards is needed. It's an enjoyable hobby project that's taking me somewhere and I cannot be certain about the final result. I built it for fun and seeing what it can do. It may not solve any problems. I still think the primary objective is to bring some life into the brain, though the exact type and nature of this machine life remains undecided. I also want to try some new things with it, like enhancements, new algorithms, new interfacing, new communications, and forthcoming ideas. I originally began building this brain for an advanced humanoid robot so that's the end result. For now, I want to improve it, test drive it, and see where it can take us..
The question was asked about the type of ANN the Brain is using. The Brain is running a sample neuron package called Neural Material or Neural Package. This is really a unique demonstration of a Neuron
Primative. What is it? What can it do? What purpose does it serve? The purpose of this demonstration Neuron is to infiltrate the Brain to a maximum density using the Neural Injectors. The next purpose is to fire each Neuron with a periocity of specificity. The next purpose is to exemplify that neural "reaction." Undoubtedly additional parameters are being explored in these early primatives.
-Phil
Brain downtime is incurred as the engineering computer containing all the Brain software has gone bad. It was apparently hit with a virus. Semantic Anti-Virus was disabled and would only scan and report virus. Another Virus program was installed and it deleted 27 virus and quarantined 56. Unfortumately, Windows explorer.exe and EXPLORER.EXE was deleted. Upon reinstall, the desktop returned but the OS would never work after that. More virus was identified that the antivirus program did not delete including startup.exe which was a hidden file Trojan-Worm. Several computer techs have worked on this for three days, are working on it today and last I heard the entire Windows OS was corrupt and will need to be installed. A reinstall is scheduled for today. It is currently unknown if the hard drive and nearly half a million files have survived the ordeal. Let's hope for the best. As the computer is five years old, it's time to replace it. Any ideas for replacing it with Mac which may be a little more virus resistant? Or maybe too many programs need PC. I am also thinking about a dual OS Mac with OSX and Windows. Any advice is appreciated.
Considering the importance of the Brain project and the immensely invaluable functions that a supporting computer must provide, and the general inability to tolerate down time, a new consideration is in effect for providing a new dedicated Brain support computer.
The dedicated Brain Support Computer will have the following requirements:
The actual Brain signal divide points where the number of Propeller chips begin to see a degradation in signal quality during parallel loading is at 56 props. This divide point is noticed at loading when the number of 56 is greatly exceeded. Additional signal conditioning may easily remedy this, though a partitioning scheme is currently chosen to immediately assure high signal quality. Tests will be run at a later date with more signal conditioning. Tests show that once the code is loaded into multiple partitions, props are free to exchange data on a net. 17x3+2+3=56 Since each partition is "driven," the number of partitions is unlimited or based on whatever is manageable. Remember that current configurations of three partitions up to 50 props each will deliver 150,000 packets of neural material.
To meet all requirements for a dedicated portable Brain Support Computer DBSC, a loaded EEE PC with the ATOM processor is chosen. Tests are now in effect to determine the efficability of massive external TB USB drives for program storage, and to pace the computer and perform setup of software. Brain code is nontransferable until the original supporting pc is repaired and returned.
It is noted in tests that the signal condition as determined by the Brain Signal Divide Point BSDP, is affected and regulated by the broadband signal of the breadboard. Some breadboards sustain and carry a more clear signal. This assessment is made with a variety of solderless breadboards tested from the following sources:
- usa
- china
- taiwan
More tests are needed to ascertain the potential to increase the BSDP in the future.A subjective analysis of the above mentioned solderless breadboards is as follows:
usa - 80% of signal
taiwan - 90 % of signal
china - 60% of signal
Signal testing is setup with arrays of Propellers to determine go/no go situations. When the original no-go situation was encounted, the number of props was backed off for the establishment of a reliable partition. The partition was built with a combination of breadboard sources.
Bad news - the Chinese MIS computer man just reported in with the status of my computer. He says somehow, when he was working on it, all files were deleted off the hard drive when "repairing Windows." He regrets this unfortunate mishap. This includes the deletion of the last 5 years of my work, including all Propeller Brain files! Hard drive inspection indicates it is cleaned. He removed the hard drive and it will be taken to another Chinese computer company tomorrow morning, apparently to see if they can find any files. I'm not optimistic about the outcome of this situation and will immediately give up all computers with Windows operating systems.
Humanoido, sorry to hear about your data loss. I've found uploading copies of source code to the forum is a free way to backup your code. I do that and upload my source code to Gmail as an attachment. Something to consider for the future.
At least You have much of Your notes about the Big Brain project right here on this forum...:cool:
What about the system you used post the last couple of threads? You can limp along with that until you get a new box.
Humanoido, I am sorry for your loss. I would very much like to have seen your "Neural Material" code. What unfortunate timing.
-Phil
I posted the previous recent posts with an Apple iPhone. I'm now running eee pc for a short time.
Unless the virus deleted files... this morning he is saying the file names do not appear and wants to know the exact names of the files on the hard drive. My response is immediately return my computer to me. It was one thing to give permission to extract a specific Windows virus but another thing, not approved, to reinstall Windows and mess with the hard drive. The drive had 50GB capacity with (9GB free) approximately half a million files. Internet here is very slow, and most sites are blocked so any kind of online storage is not practical.
I also use a live boot LINUX CD, it's a bit older than the Knoppix but does the same thing. Apparently something happened to all the Windows files and data - this is what the service man is reporting today. My guess is he reinstalled Windows.
I keep all my development data and files on the desktop. I am told now that the hard drive shows 35gb on it but 11MB is missing and there are no file names for any data. For example, thousands of .doc files don't show up.
Latest update: I'm uable to get my computer back today. The language translator said the MIS man removed the hard drive and took it home and is working on it. The company CEO is now involved and put enormous pressure on the MIS man.
The company wants to give the man time today as an opportunity to work on it. There will be an update reported today in two hours. I think tomorrow is going to be a day of reconing.
I have left messages to have the hard drive put back into the computer and return the computer to me immediately. I will work on it myself but it may take 2 days to have the computer delivered.
So which "publisher" has committed you to an NDA on their Spin code? Or is that tidbit confidential, too?
-Phil
Good points Ttailspin, As machines become more and more complex, the potential is born for more and more failures. Sometimes a machine becomes so complex, it defies troubleshooting. Many people have described Window'ed computers this way, not understanding the erratic and errant, often times intermittent strange behaviors.
One example, other than the Brain, is a complex humanoid that I was working on. I had the surprise of my life when I brought it to life for an extended period of time (charging and recharging batteries), then let the battery run down - big mistake! The humanoid flung out its right arm and hand then brought it back to the chest, as it it was having a heart attack.. it stumbled forward a few steps then fell limp to the floor in a twitching leg and full body spasm that was extremely disconcerting and worrisome to see. After that happened, I took measures NOT to let any humanoids die. Even in the BASIC Stamp SEED Supercomputer I wrote code so the ten life forms would not die. Unfortunately the Brain's supporting computer bit the dust before I could replace it.
Good point, I'm happy with the number of posts and the volumes of information made available on the Forum about the Brain and after I get my new computer I can address the possibility of reconstructing the software over time.
My only experience with The Knoppix CD has been to restore files from an (at least partially) intact file system. If your file system has been completely rebuilt, the Knoppix CD may not be of much use.
___________________
So let me get this straight. You've signed an NDA with a "publisher" who provided Spin code for some kind of neural simulation. Am I right so far? Or did they provide the algorithm, and you converted it to Spin? And, just to clarify, this "publisher" won't even let you say who they are? Do I have that part right as well? Also, what changes in a "couple months" that frees you from this NDA to publish raw code?
-Phil
-Phil