Відео

Thank you for being interested! I am working on the next video now. I decided to make it about the fundamentals of neural networks. I am hoping that I will get it done by Monday. I really want to get the hardware-based neural network video done but going over fundaments is important.

The light would add some charge to the receiving side. If we use the optocouplers that are in the IC packaging the outside light and light from other synapses would be blocked. It is a good point though that if we used a sensitive receiver in an open environment like this it could add charge to other synapses. That could actually be a design feature rather than an issue if we wanted one output to contact many other inputs without having to wire them all together.

@@GlobalScienceNetwork Or allow a limited number of other neuron inputs to be coupled to a single (or more) neuron on a computational level. The problem as I see it is being able to get adjustable light signal levels from each input independently with minimal circuitry. :)

Ha ha thanks! I figured if I was giving a talk to hundreds of people I would wear a suit. So I try to make the same effort for UA-cam content.

I am glad you found the video helpful! I am working on the next video right now! There should be some cool projects coming up soon. Thanks for following along!

I am glad you found them helpful!

That is great you are learning about these types of circuits!

Not from the wires or components on the breadboard. If I powered the computer with a USB cable from the wall and an unshielded cable there was some noise on the incoming signal. Since this was a digital computer the noise was not enough to make the transistors not work. It is something to be careful of though when making analog designs.

Sure thing, glad you found it helpful!

I am glad you were able to find the video and now understand how these circuits work!

Ha ha when I took my first programming class in HS I was surprised. I thought this was not actual programming, this is typing in a program someone else programmed. I mean let's at least start with machine code.

Because it's cheaper than the 2n2222 transistor

​@@divyaswarupmishra5388 Those should work. I am pretty sure the orientation is backward though, the emitter-base collector is the collector base-emitter. This just means you have to rotate the transistor 180 degrees.

Now sure why I would be the NPC.

Ha ha that would be good goal!

Most of them will work but the BB830 by busboards are the best and worth the money if you have the funds! This was built with average quality boards and worked but I did have to swap about 4 of them out that had issues. So now I use the better breadboards.

Well I build a calculator using full adders in a different video. We also used counters to build a full computer in a differnt video. The actual operation of this counter is division which could be added to the ALU of a computer.

ha ha

Glad you found the information helpful!

Yeah , I should have recoded that and show the faster rate. I did not test the max clock speed but it should actually run to around 200,000 Hz.

Good question, as current flows into the base of the transistor the transistor allows current to pass with some but a very small amount of resistance (around 8 Ohms). So now the current going to ground has to go through two resistors. Since the 2K resistor is so much larger almost all the voltage drop occurs across that resistor. Does that make sense to you?

​@@GlobalScienceNetworkyeah...now I understand better but still I don't know how it was calculated as 4.984v through 2k resistor and 0.16v through LED. Could you please mention the way to calculate these!!

@@Mani-wq1de Well how much equivalent resistance there is across a transistor is not really a value that is common knowledge. Once you have it though the voltage drop is calculated as two resistors in series. I got it from a circuit simulator. The .16V is what would be left. This is not enough to go through the LED as a yellow LED requires a voltage drop of 1.9V to turn on. I was pointing out that this is why the LED is off.

Sweet, sounds like a great setup! Building circuit projects is really fun! There are lots of possibilities!

Well, you brought up a lot of points so I will try and respond to each. 1) It would be cool to re-use synapses, overall this is a good idea. This is not how the brain works, so it depends on how closely our final design matches the brain. If the hardware is hard-wired it would have to be built into a design like an FPGA to change wire locations or which synapse is used. 2) Microcontrollers are cool but is this not the same as running a simulation on a digital computer? The idea is for this to be an analog neural network-style design. 3) Functionality is the biggest problem. How many neurons are needed to do what, as a test? This is a great point. We need to build a verified model/test case to work towards. I am still thinking about what would be a good simple test case. 4) GPT can't think because it lacks a memory. This is an interesting point. I am actually outlining all the ways in which simulated neural networks fall short of being functionally equivalent to biological neurons. Basically, neural networks are trained to create a predictive model. Then when given an input it can predict a good output. As you mentioned though the model does more really store memory, it also does not process information in a way so that it can have consciousness.

ha ha thanks!

Your wavelength argument seems solid. That is ok if you do not subscribe. Check out my most recent video's about building artificial synapses and neurons with your app. I bet you will enjoy them!

Ha ha yeah I like building the hardware though! If you build hardware properly you actually should not need any coding.

Yeah, it is difficult to build a circuits that are comparable to the human body/brain, that is why we have not done it yet. Neuroscientist are sill trying to map all the connections in the brain and they have a long way to go. It is the only science project worth working on in my opinion though as the results have real meaning.

These logic gates are build with 2N2222 and 2N2904 NPN transistors. Either one can be used for all of these logic gates.

Yeah, a microcontroller is a good way to implement most robotic projects quickly. It will be interesting to see what we can build using artificial neurons!

Yeah, it would take a couple million breadboards ha ha!

Thanks for the sub! If it was 8-bit it could be interfaced with software to run simple programs.

Yeah, that is the plan for future videos! The next video is going to be how to build neural networks. What type of input and output devices we should use is an important question. Once we get a design refined on breadboards we can have the design made on PCBs which should make it easier to integrate with lots of artificial neurons and should be flexible for different input and output devices.

Ha ha thank you!

Thank you! Yes, adding the 100K resistor would be a good addition to the circuit! I am happy to see that electrical engineers are following along and providing feedback!

Thank you! I have some more great videos planned as well! Keep following along as things are going to get awesome!

Glad it was helpful, good luck with your studies and future electronics projects!

Great thanks, keep following along and provide feedback when you can. I just uploaded a new video about building artificial neurons.

@@GlobalScienceNetwork Link please

@@ufehboss2566 ua-cam.com/video/Uhuo9ketdhg/v-deo.html

You should be able to make a full adder with a BC547. Maybe you have a wire wrong or a bad transistor. Also realize the transistors would have to be flipped as the emitter-base collector for a 2N2222 is the collector base emitter for the BC547 if it was in the same orientation.

Well said GlevoTec. When I was learning electronics, I had to build circuits to understand the theories and concepts that I was learning in class. Luckily, my father was an electronic engineer, so there was never a shortage of components around my house. Learning electronics would have been MUCH more difficult for me if I had to do it without hardware. I've used Arduino boards to make projects where a microprocessor is necessary for the functionality of what I want to do. I've also been amazed by the complexity of the design and programming that some go through to blink an LED with a microprocessor. That's something I could do with a 555 timer or a few transistors, a couple of passive components, and a 9 V battery.

@GlevoTec thanks for the response. I am planning to build basic intelligence into a robot as well using artificial neurons for the logic. I am curious what input sensors you think would be best to use for vision. Do you know of a camera where a grid output can be easily obtained. Or just use individual IR Infrared Obstacle Avoidance Sensor or ultrasonic sensors. I have the ELEGOO Smart Robot Car Kit V4 and I am probably goin to see if I can interface some of these sensors with the neurons unless I come up with a better idea.

@@GlobalScienceNetwork Thanks for the reply. I do have some ideas for vision actually. I've sent you an e-mail few days ago with some questions. I'd be happy to discuss some ideas there too. Greetings

@@GlevoTec I replied to your email now. For some reason it went to span so I did not see it. Thanks for the message!

I have some more videos coming soon that will get into more detail. Consciousness may seem a bit obscure but hopefully I can explain it in a simple way. If we build circuits properly we should be able to create non biological life.

@@GlobalScienceNetwork I don't think consciousness is obscure. It is the most common "thing" we know as conscious beings. I think it is clear that consciousness is that, in which all Qualia appear, i.e. that in which all things appear, but it is not itself a thing. That means, it is "Being" itself (with a capital "B"). My experience is not that I "have" consciousness, but that I "am" consciousness. My body appears to me in consciousness. If I had no body, I could imagine to still be. But if there was no consciousness, there would be nothing left. I believe the idea of consciousness being something that can be "created" is a gross misunderstanding of reality. By the way, if you build a circuit, how would you decide if it is conscious or not? How do you know it does not only behave like a conscious being, but actually has inner experience? I mean, I am a human being, but I'm totally unable to proof to anyone else that I am conscious myself (experience Qualia).

@@valuemastery Well there are lots of tests to determine if something is conscious by the classical definition such as the Glasgow Coma Scale or that it can self-identify or be empathetic towards others. Much of empathy is mirror neurons allowing us to live experiences of others just by watching them. I will explain a more simple definition with different levels of consciousness. In biology, there is also the 7 characteristics of living things such as the use energy, can grow, respond to stimuli, maintain homeostasis, reproduce, adapt, and is made of cells. Sure we will be creating a device/life that will be consciousness not just creating consciousness if that is what you thought was a gross misunderstanding of reality. Maybe you think simple life forms are not conscious however I think they are, it is just a lower level of conscience and as life gets more complex the level of consciousness increases. The platform or space for consciousness to exist does need to be continuous with direct interaction with the environment. This means that it would be nearly impossible to create a conscious being within a digital computer. However, it should be relatively simple with hardware-based neurons.

@@GlobalScienceNetwork I do build neural networks myself, as well as FPGA based computers. So I think I have some understanding of hardware and artificial neurons. From my point of view, all they do is create a processing network. What should be the criteria for it to be conscious? All definitions we have are behaviour based. Even integrated information theory assumes that for creating consciousness the basic building blocks need to carry some level of consciousness to start with, so with increasing complexity of information processing the level of consciousness would increase as well. But I think the error lies in believing that neurons or switching devices, or anything for that matter, carries consciousness at all. That worldview is known as panpsychism. If it is true (which I don't think it is), it would mean that consciousness can't be created, just aggregated. However, I think it would be more logical to view consciousness as primary, since it is much closer to our experience. So we have it backwards: Physicality does not create consciousness, but consciousness is aware of physicality. So the physical world is just an appearance of consciousness within itself. This would also mean that there are not many consciousnesses, but only one; and the seemingly separate conscious beings are just disassociations within consciousness. From what you say it's clear that you have the opposite view. However, I strongly feel this is wrong, hence I don't believe that any physical process can create consciousness. I still have to meet a scientist who could explain how this could work, even theoretically.

Thanks!

No problem, sounds like you got it figured out!

Thanks, that is the plan! I have the summer off now and am almost finished moving so I will be getting some more content up soon!

Yeah near the beginning of the video I show the full logic gate level circuit diagram. It is in 4K so you should be able to read the details. If you email me I can also send it to you.

Thank you!

Thanks! The goal was to build a computer at the transistor level to ensure we know the fundamentals of how a computer works. Now I am working on building hardware-based neural networks. The binary to seven-segment display would have been a good addition. The binary numbers are pretty low though so reading the binary is pretty straightforward.

I agree 100 percent! If you follow along I bet you will be able to help improve these designs!

Great, let me know if you have any questions!

I use a 5V 2 amp battery back. You can even use a 5V cell phone charger. As you say a 5v regulated power supply would work as well but the other two are simpler for most people. You can make a custom wire or use and breadboard adapter and power from a USB cable.

Sure thing, thanks for watching!

Yeah, that would be a fun project! With some of the artificial neuron circuits, I plan to make them PCB projects!