- 19
- 426 034
Global Science Network
United States
เข้าร่วมเมื่อ 17 พ.ค. 2021
At the Global Science Network, we focus on three main projects which are solving the world's energy problems, solving unified field theory, and creating non-biological human consciousness. With your help, we can work on these projects in a highly collaborative way.
Neural Networks Explained: Solving the XOR Logic Gate with Backpropagation
The XOR logic gate is an excellent example of a neural network that can be trained and utilized to solve complex problems. A non-linear decision boundary is needed as the XOR logic gate is not linearly separable. This means that a multi-layer neural network is needed.
Backpropagation will be used to train the network, and how this is done will be shown with an architecture diagram and with the equations. A numerical example will be provided to show how the equations are solved.
To train the network, 40,000 epochs are run to optimize the weights and biases. The network was trained using both stochastic and batch gradient descent, which resulted in similar solutions.
Adding noise to the input points in conjunction with adding a hidden layer made it so the resulting decision boundary plot was very close to the ideal case.
Backpropagation will be used to train the network, and how this is done will be shown with an architecture diagram and with the equations. A numerical example will be provided to show how the equations are solved.
To train the network, 40,000 epochs are run to optimize the weights and biases. The network was trained using both stochastic and batch gradient descent, which resulted in similar solutions.
Adding noise to the input points in conjunction with adding a hidden layer made it so the resulting decision boundary plot was very close to the ideal case.
มุมมอง: 621
วีดีโอ
Neural Networks, From First Principals, A Single Perceptron.
มุมมอง 938หลายเดือนก่อน
Understanding the fundamentals of neural networks is very important. For this reason, understanding how a single perception works is a good place to start. A single perception can separate two linearly separable classes in 2D or 3D. This is done but using the perception learning algorithm, which is also called the perception learning rule, or just the delta rule. When a given x1 and x1 input is...
How to Build Artificial Neurons
มุมมอง 6K5 หลายเดือนก่อน
Artificial neurons could be the future of life in our universe. This video shows how to build actual artificial neurons on breadboards. Artificial neurons can be used to build hardware-based neural networks which when properly implemented could turn into actual independent lifeforms. Lots of work needs to be done to get to that point but the future is very exciting for this type of technology. ...
How to Build an Artificial Synapse
มุมมอง 14K10 หลายเดือนก่อน
Artificial synapses can be built with basic electronic components on breadboards. The artificial synapses are made with an optocoupler, diode, and a resistor. Transistors are used to buffer the input and output signals. The transmitting side of the optocoupler can be built with an LED or an IR LED. The receiving side of the optocoupler can be built with LED, photodiode, phototransistor, or ligh...
How to Build a 4-Bit Computer on Breadboards Using Individual Transistors
มุมมอง 193Kปีที่แล้ว
A simple 4-bit computer is built on breadboards using individual transistors. This computer is called the GSN477. This is because it was built for the Global Science Network, is 4-bit, and contains two 7-stage counters. Sections of the computer are the clock, program counter, ring counter, 10 bytes of memory, opcode register, opcode decoder, control matrix, data bus, accumulator register, outpu...
How to Test Transistors
มุมมอง 3.2Kปีที่แล้ว
Testing transistors is important before building circuits on breadboards. The test shown is just a basic test and does not characterize the transistor. A NAND gate is built and when the transistor is inserted it should turn the LED off. The transistor is wiggled around and the LED should not flicker. If it does I would not recommend using the transistor. Testing the transistor takes time but sa...
How to Build a Computer on Breadboards Using Transistors
มุมมอง 13Kปีที่แล้ว
How to build a 4-bit computer using individual transistors will be demonstrated and discussed in detail. Parts of the computer include the clock, memory, accumulator register, register B, ALU, and output register. An example program will be run where load A, add, subtract, output and halt are executed. The computer will be put in a loop where it will continuously add by two and then subtract by...
How To Use LTspice, A Free Circuit Simulator
มุมมอง 16Kปีที่แล้ว
This tutorial shows how to use LTspice, which is a powerful, open-source circuit simulator. It starts out by drawing a simple circuit with a voltage source, resistor, LED, and ground. A custom model is made for the LED. Next, a full adder is built at the transistor level using NAND logic gates. The full adder is turned into a symbol so a 4-bit calculator can be stimulated. Finally, an astable m...
How To Use EasyEDA, A Free Circuit Simulator
มุมมอง 28Kปีที่แล้ว
How to build and simulate circuits with EasyEDA is discussed in detail. First, a basic resistor LED circuit is built and the voltage and current values are measured with a multimeter. Second, a 4-bit calculator circuit is simulated. Finally, an Astable Multivibrator is simulated for ten seconds and the output is shown on a logic analyzer and on an oscilloscope. 0:00 Intro 0:38 How to launch the...
LED Current Limiting Resistor Values
มุมมอง 8Kปีที่แล้ว
Using a current-limiting resistor is important when using LEDs. If the resistance of a circuit is too low the high current can limit the life of the LED or burn it out almost immediately. The goal of an LED is not always the maximum amount of light and a properly sized resistor can help achieve the proper brightness. 0:00 Intro 0:24 Voltage drop of diode vs LED 1:15 Voltage drop of different co...
Master Slave JK Flip Flop and Data Flip Flop
มุมมอง 2.8Kปีที่แล้ว
How to build a master-slave JK flip-flop and data flip-flop is discussed in detail. This was done on breadboards using individual transistors. Each Flip flop is built with NAND logic gates. The clock used is an astable multivibrator. 0:00 Intro 0:25 Edge Triggered JK Flip Flop 1:13 Master Slave JK Flip Flop 2:31 Circuit Design Explanation 7:56 Clock and Output Timing Diagram 9:14 Ouptus on the ...
D Flip Flop and JK Flip Flop
มุมมอง 6Kปีที่แล้ว
Data Flip Flop and JK Flip Flop 0:00 Intro 0:30 Data Flip Flop vs Data Latch 1:05 Data Flip Flop 2:40 Trigger 1, (using AND gate and inverters) 5:21 Trigger 2 (using capacitor and resistor) 9:22 JK Flip Flop 11:33 JK Flip Flop with edge-triggered clock How to build a data flip-flop and a JK flip-flop using individual transistors will be discussed in detail. The flip flops are triggered by a clo...
SR Latch, Gated SR Latch, and Data Latch
มุมมอง 7Kปีที่แล้ว
In this video, I show how to build an SR latch, a gated SR latch, and a data latch using individual transistors. 0:00 Intro 0:29 Two Transistor SR Latch 1:04 OR Gate Latch 2:03 NOR Gate SR Latch 4:52 NAND Gate SR Latch 6:52 Gated SR Latch 9:27 Data Latch 11:18 Data Latch vs Data Flip Flop
Binary Counter
มุมมอง 9Kปีที่แล้ว
This 4-bit binary counter is part of a 4-bit binary computer that I am building on breadboards using individual transistors. An astable multivibrator is used as the clock of the computer and counter. The first bit of the counter is the output of the astable multivibrator. For the second and third bits, an edge trigger data flip flop is used. Each trigger is made with an AND gate and 7 inverters...
4 bit calculator built using individual transistors
มุมมอง 8K2 ปีที่แล้ว
A 4-bit calculator is built using different types of logic gates for each full adder. It is able to add up to 31. The inputs and outputs are shown by lighting up LEDs. The first full adder is built with integrated circuits that consist of XOR gates, AND gates, and OR gates. Next, the second full adder is built with individual transistors using the same logic gates as full adder 1. The third ful...
Learn how computers add numbers, 4 bit adder circuit
มุมมอง 6K2 ปีที่แล้ว
Learn how computers add numbers, 4 bit adder circuit
Digital Logic Gates from Transistors, AND, NAND, OR, NOR, XOR, XNOR, Buffer, and Inverter
มุมมอง 85K2 ปีที่แล้ว
Digital Logic Gates from Transistors, AND, NAND, OR, NOR, XOR, XNOR, Buffer, and Inverter
It's a little bit monotonic, but after some time, everything becomes clear
Good deal!
Tobey Maguire looking good
Ha ha thanks! I have had a few people say something similar!
I found it very non-intuitive! too many steps, different pages, pro and standard, too complicated to find a component, and I couldn't place it on the design page! I gave up
Just make sure you are in standard edition, simulation mode. Then it should work as expected. That I why I was made the video to try and help. I have another video about LTspice if you prefer that circuit simulator.
at 0:43, why do we have 3 transistors for the AND gate? Is it because when we connect output at the collector, and lets say both inputs are on, the current will just go to ground and wont reach the output, which is NAND gate, but then that 3rd transistor acts as an NOT gate which negates the NAND, making it AND gate, if that is the case, then why bother connecting output at the collector and not just go straight to emitter?
This is a good though. You could try it and see if it works. My guess is that connecting it to the collector will not allow the output signal to be sent to the next circuit.
Now I understand resistors, thank you.
Good deal! Thanks for watching!
@GlobalScienceNetwork - Sir, thanks to you, I hauled myself down to an electronics shop today and have just returned with a whole bag full of components to play with. Happy New Year greetings from Cape Town.
@@friendlypiranha774 Sweet, Happy New Year! That is great there is an electronic shop around as well!
Yeah but can it run pong?
We would need to add some more transistors for that! ha ha
@@GlobalScienceNetwork So, when's part two 😛
Hello :) Thank you a lot for the efforts put in all of your videos ! You're incredibly good at teaching, and all your explanations are accompanied with good drawings too. You are a very inspiring person, and I am now interested in little electronics :) Sadly I'm not good (yet !) to fully understand everything... but I did buy 300 2n3904 transistors and 200 2k resistors to try and learn the basics. (also I will start a robotic club at my school, I do some 3D printing from time to time ^^) As for the learning aspect of neurons, I was thinking... (probably not a solution since I don't understand everything) but what if we had capacitors that could change their capacity based on some voltage input ? I don't know how such capacitor would work, but let's assume that we have such capacitor (we'll call it a voltage variable capacitor = VVC) (actually, as I was searching online, I found this: www.eeeguide.com/voltage-variable-capacitors/ but I don't understand all of it) the VVC would have 2 input and 1 output: - the first input is the charge it receives (just like a normal capacitor) - the second input will somehow change the capacity of the VVC. so if it receives 1V, it will have 10 pF, 2V 20pF... - the third is the discharge (when needed, if we assume it discharges after a certain threshold, then it could discharge more often if it had a bigger capacity ?) Since I don't understand everything about electronic, I'm not sure it makes sense, but I guess the higher the frequency of the signal the synapse receives, the bigger this VVC capacity should be (hence, it would be able to receives more... uh... Energy ? again, I probably don't make a lot of sense since I myself don't understand how I would implement that circuit) Anyway, thanks a lot for this video ^^ I'll let you know how my club goes, and I wish you the best in your endeavour :)
Why can't the OR1 gate send output to other logic gates, but OR2 can?
OR1 completes the circuit to make the logic true as a loop within the circuit. That will only be useful as a stand alone logic gate. To send it out to other logic gates you basically have to invert that logic and one more inverter transistor is needed.
Thanks teach I needed this. Loved your algebra class btw
Great! Thanks for watching!
everyone should know how to build a computer on a transistor level 🤣🤣
ha ha well we all need to know how to read and do math so this would just be the next step! At least every electrical engineer and computer science major should now how to build a computer at the transistor level. It should be in programming class 101.
@GlobalScienceNetwork chill man I'm an electronic major and I'm just kidding 😅
@@nehal-z8y Yeah, I know. When I took my first programing class this is really what I was hoping to be learning though. It was a GW BASIC class in high school and I was thinking I wanted to learn how to write the GW BASIC program not just use someone else's program.
Hey! I absolutely love your videos, but I have a tutorial request: I cannot figure out how to use logic ICs of the 74HC series. Can you please make a tutorial about that?
Hi, at the end of the logic gate video I do show how to wire them. One way I call the wrong way but in hindsight either way will work. I also wire some ICs up for the 4-bit calculator built with differnt types of full adders.
Is it turing complete? And its very cool🎉
For a machine to be Turing complete, it needs: 1) Ability to read/write memory. 2) Have conditional branching. 3) The ability to perform basic arithmetic operations. 4) Theoretically unlimited memory (though physical implementations always have limits). 1) It can read memory and temporarily store results in the output register. 2) It can run differnt commands but does not have jumps which could be added easily. 3) The computer can add and subtract. 4) Memory is very very limited but theoretically could be expanded. So yes I would say that it is Turing complete.
Hello, very cool
Thanks!
تحيا لك من ليبيا🇱🇾❤️
شكرا!
Note: If you go to settings, audio track you can listen to this video in 6 different languages (English, Spanish, Portuguese, German, French, and Italian). This is a nice feature TH-cam Recently added. If your language is not there, sorry, maybe TH-cam will add more languages as this is a new feature. As far as I know, this will be done on all future videos but not previous videos.
it has come full circle!
Almost! Building hardware-based neural networks is the goal. So I plan to make one more video showing how to program neural networks than the moment I have been waiting for. Implementing and improving hardware-based neural networks. It is going to be awesome!
Hi! In the future, can you post links to previous videos as well? Thank you!
If you click the channel you can see all the videos including the latest videos.
How doeas the inverter works? Can you explain it to me? I got confused :(
Well the inverter makes an on input signal into and off input signal, and an off input signal into and on input signal. The circuit has one input and one output. I would recommend trying to build the circuit and testing it out. If you have a specific question about the circuit, let me know.
@GlobalScienceNetwork Okay, thanks
What I dont understand in the BUFFER2 is that, why the eletricity travels trough the transistor (if it's on) but not trough that blue wire? I mean that would be a shorter way, wouldn't it? Sorry if it's a dumb question...
thank you for making such a wonderful, amazing, informative video....lots of love from india🇮🇳
Sure thing, good luck with your circuit projects!
What is the purpose of opto-coupler why not a simple direct connection?
In this case the optocoupler is basically is a buffer. However the optocoupler could be modified to make it so the amount of charge transferred varies based on the amount of light and wavelength of light. The could be changed in a neuromorphic way based on the activity of the circuit.
Damn bro didn't even use 7400 IC chips, this some low low level stuff
Yeah, I am trying to show how these things work at the transistor level!
Could you please explain why in OR1 you connected transistors' collectors, but in NOR gate you connected their bases? I'm a bit confused. Also, I always thought the right and left pins are emitter and collector respectively but in this video it seems it's opposite. Sorry I'm new to all of this but I love it. And thank you for your work!
The inputs should go into the base for all of the logic gates I made in this video. The extra resistor on the collector allows the output signal to be sent to other circuits. It is not an on or off for the circuit. There are logic gate designs were this is not the case. For the 2N2222 and 2N3904 it goes emitter(left) base(middle) and Collector(right). Depending on the model the emitter and collector orientation can be switched.
great
Thanks!
That's so amazing. I simulated a 4 Bit Adder already, and this is logicially the next step, building it. You can't imagine how nice it is to have a reference video, i really don't understand why not more people watch your content. Anyways, thanks a lot!
Awesome! I also think that building the 4-bit adder is a fun project! Let me know if you have any questions as you build other circuits.
@@GlobalScienceNetwork I'm sure i'll have fun. I have electronics and digital logic as subjects in school, but i've never been taught how to build stuff using (almost) only transisors. I think this will help my understanding of computers grow, which is my goal ultimately.
@@captainvaughn5692 Great, yeah I think it will!
Excellent. I can learn those now by taking notes from this video. I have a delivery that is a breadboard and it's coming soon. I'll rate this 10 star ⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐
Thank you! Great, good luck with your circuit projects!
How did you reply so fast?
@@rechieandramsey8 I am on my computer and I get notifications.
Oh😅
Also, are you always in your computer?
I'm going to make one :D it is amazing
Awesome, let me know if you have any questions!
Hi, making this artifical neurons can help blind people to see and much more..... Lets work on this...... I am researching to make an electronic device that can communicate with neurons connected with the human vision system. This device aims to utilize understanding of how electrical signals are transmitted within the nervous system. While it won't directly communicate with neurons, it will play a crucial role in the physiology of neurons. Neurons rely on the movement of ions such as sodium, potassium, and calcium across their cell membranes to generate electrical impulses, creating gradients for signal transmission along the neuron's axon. The device will likely interface with these ion channels and receptors, influencing the electrical activity of neurons. Through this interaction, it could potentially enhance or modify visual perception, offering new possibilities for assisting individuals with vision impairment. If the neurons in existing vision systems are not producing electrical signals or are unable to function properly, there are potential strategies to artificially activate them using external devices. One approach involves using technologies such as optogenetics or electrical stimulation to directly stimulate the neurons and induce electrical activity. Optogenetics involves genetically modifying neurons to respond to light, allowing for precise control over their activity using light pulses. Electrical stimulation, on the other hand, involves applying electrical currents to neural tissue to elicit neuronal firing. By bypassing the malfunctioning neural pathways and directly stimulating the neurons, it may be possible to restore or enhance visual function in individuals with vision impairment. However, developing such technologies requires a deep understanding of the underlying neurophysiology and careful consideration of ethical and safety implications. Additionally, further research and testing would be needed to assess the effectiveness and potential side effects of these approaches before they could be implemented clinically. Creating a tiny device with a battery that can trigger nerves involves several challenges but is theoretically possible. Here's a simplified overview of what would be involved: 1. **Miniaturization:** The device would need to be extremely small to be implanted near nerves. This requires advanced microfabrication techniques to create components on a tiny scale. 2. **Power Source:** The device would need a compact and long-lasting power source, such as a miniaturized battery or energy harvesting system that can convert ambient energy into electrical power. 3. **Stimulation Method:** The device would need to deliver precise electrical stimulation to the nerves. This could involve using microelectrodes or other stimulation methods compatible with the small size of the device. 4. **Control System:** The device would require a control system to regulate stimulation parameters, such as frequency and intensity, to effectively trigger nerve activity. 5. **Biocompatibility:** Since the device would be implanted near nerves, it must be biocompatible to avoid adverse reactions or tissue damage. 6. **Safety and Regulation:** Ensuring the safety of the device is critical, requiring rigorous testing to assess its effects on neural tissue and regulatory approval before clinical use. Developing such a device would require interdisciplinary collaboration among engineers, neuroscientists, and medical professionals. It would also involve extensive testing in animal models and potentially clinical trials to evaluate its safety and efficacy in humans.
Where is the screen? How to play flappy bird on this computer? And where to store data? In a wooden box? 😊
Good questions, the program is stored in the memory. The output is displayed in the output register at the end of the program. There is not a save feature for the output. The screen would be a good addition but right now it is just the LEDs showing the binary values. We would need to add a couple million/billion more transistors to play flappy bird!
@@GlobalScienceNetwork million/billion more transistors? Does the current PC or Laptop have that many transistors? Or do you mean you need a CPU to hold that large a number?
@@Al3abmizo Yeah, there are billions in the CPU or microprocessor. There are also billions in solid-state memory, GPU, and RAM.
This video deserves a lot more views. It is educational and amazing to see someone stimulating nerves with only eletronics components you can easily buy!
Thank you! Hopefully with time it does get more views! Once people see what we can build with hardware based neural networks the interest in the content should increase!
Keyword: analogue computer. They are even older than binary systems. before anyone wonders the binary/digital systems have prevailed because they are extremely flexible due to the way they are programmed.
Yeah, that is true. It is likely that hardware based neural networks and neuromorphic computing will be a big part of the future though. Digital computing wont go away but we will get better at building analog computers.
two feelings in my heart: envy and respect.
Ha ha Thank you! If you want to build one and have any questions, let me know.
Make the robobrain run doom.
Ha ha, I will very likely try and make that a project!
can we make a 1 bit adder by only npn transistors
Well with only 1 bit what would you be adding? A full adder adds two bits. If you mean a 1-bit register, that is shown in the full computer build.
my first video ... looks to be a very good one ... (for whatever strange reason he pronounces emitter as 'eDmitter' 😝...
Ha ha that seems like a valid point! Check out the other video's as well. We use these logic gates to build a simple computer. Right now we are working on hardware based neural networks.
finally caught up to the most recent video. This has been a pretty exciting series.
Thanks, more videos coming soon!
Not sure how you managed to get 5v dc to flow through your body……
Yeah, it is pretty wild! You are a conductor so in this case your body is acting like a wire.
Can you share the circuit diagram? And components rating please.
I show the transistor level circuit diagram in the video. It is in 4K so you can just screen capture it. Is there a rating of a specific component you are wondering about?
Can i know the circuit connection diagram.
The logic gate level diagram is shown toward the beginning of this video. When I show the individual sections I typically show the transistor level circuit diagrams.
This is really interesting, thanks for taking the time to upload this. There are so many resources to simulate circuits, would you recommend EasyEDA over LTSpice? Is one more accurate than the other or is it just preference for how you use it?
OMG
Thank you!
Bro i am from Bangladesh, I need a full circuit diagram.
The transistor level circuit diagram is show for the synapse in this video. It is shown for the neuron and synapse in the how to build artificial neurons video.
But can it run Crysis?
Ha ha good question. We would need to add a couple billion more transistors!
Hey man do you have a transistor-level diagram of the computer?
I have the logic gate level diagram of the full computer. I do show the transistor level circuit diagram of each section throughout this video and my other videos.
Instructions unclear I built a windows eleven gpu
If you have any specific questions, let me know!
possibly for a 8 bit computer?
It certainly could be done! At some point I may try and make an 8-bit breadboard computer using MOSFET transistors so it will be more power efficient.
"nonbiological human consciousness" is a bit of a stretch.
Well right now it may seem that way. However, in the future we will very likely have non-biological life. How soon this happens depends on how many people and resources go to this type of project. We should make some decent progress in the next few years.
@@GlobalScienceNetwork You are taking a strong materialistic position with that statement, assuming that you are talking about phenomenological consciousness and not some arbitrary defined thing. I suggest you take a look into "the big problem of consciousness" and what the problems of materialism are. I think it has never been more obvious that materialism is on a really shaky ground and is one of the least likely ontological truths. And this is coming from philosophers, neuroscientists and quantumn physicists.
Your neck must hurt all day, carrying a brain around that size. I'm impressed!
Haha thank you!
Thanks
Sure thing, thanks for watching!
How would I go about making this smaller version of the 4 bit computer? I found the logic gate schematic for the larger one on your website but I couldn’t for this smaller one
Good question. This is just the first half of the larger computer. Basically it is the right half. You can test the computer when it is half built. So the diagram would be the right half of the larger version.
@@GlobalScienceNetworkwait wouldn’t I need the other half for this project, to get like the ring counter and stuff or is it just extra things that aren’t needed
@plantor0728 Well they are needed for a full computer but in this video, I show how you can test the first half of the computer. Each section can be built and tested separately and then combined to make the full computer.
@@GlobalScienceNetwork oh then that’s perfect then, thank you for the help
Using your body to complete the circuit is shocking!
Ha ha literally it is shocking!