This is a three. It’s sloppily written and rendered at an extremely low resolution of 28 by 28 pixels. But your brain has no trouble recognizing it as a three and I want you to take a moment to appreciate How crazy it is that brains can do this so effortlessly? I mean this this and this are also recognizable as threes, even though the specific values of each pixel is very different from one image to the next. The particular light-sensitive cells in your eye that are firing when you see this three are very different from the ones firing when you see this three. But something in that crazy smart visual cortex of yours resolves these as representing the same idea while at the same time recognizing other images as their own distinct ideas But if I told you hey sit down and write for me a program that takes in a grid of 28 by 28 pixels like this and outputs a single number between 0 and 10 telling you what it thinks the digit is Well the task goes from comically trivial to dauntingly difficult Unless you’ve been living under a rock I think I hardly need to motivate the relevance and importance of machine learning and neural networks to the present into the future But what I want to do here is show you what a neural network actually is Assuming no background and to help visualize what it’s doing not as a buzzword but as a piece of math My hope is just that you come away feeling like this structure itself is Motivated and to feel like you know what it means when you read or you hear about a neural network quote-unquote learning This video is just going to be devoted to the structure component of that and the following one is going to tackle learning What we’re going to do is put together a neural network that can learn to recognize handwritten digits This is a somewhat classic example for Introducing the topic and I’m happy to stick with the status quo here because at the end of the two videos I want to point You to a couple good resources where you can learn more and where you can download the code that does this and play with it? on your own computer There are many many variants of neural networks and in recent years There’s been sort of a boom in research towards these variants But in these two introductory videos you and I are just going to look at the simplest plain-vanilla form with no added frills This is kind of a necessary prerequisite for understanding any of the more powerful modern variants and Trust me it still has plenty of complexity for us to wrap our minds around But even in this simplest form it can learn to recognize handwritten digits Which is a pretty cool thing for a computer to be able to do. And at the same time you’ll see how it does fall short of a couple hopes that we might have for it As the name suggests neural networks are inspired by the brain, but let’s break that down What are the neurons and in what sense are they linked together? Right now when I say neuron all I want you to think about is a thing that holds a number Specifically a number between 0 & 1 it’s really not more than that For example the network starts with a bunch of neurons corresponding to each of the 28 times 28 pixels of the input image which is 784 neurons in total each one of these holds a number that represents the grayscale value of the corresponding pixel ranging from 0 for black pixels up to 1 for white pixels This number inside the neuron is called its activation and the image you might have in mind here Is that each neuron is lit up when its activation is a high number? So all of these 784 neurons make up the first layer of our network Now jumping over to the last layer this has ten neurons each representing one of the digits the activation in these neurons again some number that’s between zero and one Represents how much the system thinks that a given image? Corresponds with a given digit. There’s also a couple layers in between called the hidden layers Which for the time being? Should just be a giant question mark for how on earth this process of recognizing digits is going to be handled In this network I chose two hidden layers each one with 16 neurons and admittedly that’s kind of an arbitrary choice to be honest I chose two layers based on how I want to motivate the structure in just a moment and 16 well that was just a nice number to fit on the screen in practice There is a lot of room for experiment with a specific structure here The way the network operates activations in one layer determine the activations of the next layer And of course the heart of the network as an information processing mechanism comes down to exactly how those activations from one layer bring about activations in the next layer It’s meant to be loosely analogous to how in biological networks of neurons some groups of neurons firing cause certain others to fire Now the network I’m showing here has already been trained to recognize digits and let me show you what I mean by that It means if you feed in an image lighting up all 784 neurons of the input layer according to the brightness of each pixel in the image That pattern of activations causes some very specific pattern in the next layer Which causes some pattern in the one after it? Which finally gives some pattern in the output layer and? The brightest neuron of that output layer is the network’s choice so to speak for what digit this image represents? And before jumping into the math for how one layer influences the next or how training works? Let’s just talk about why it’s even reasonable to expect a layered structure like this to behave intelligently What are we expecting here? What is the best hope for what those middle layers might be doing? Well when you or I recognize digits we piece together various components a nine has a loop up top and a line on the right an 8 also has a loop up top, but it’s paired with another loop down low A 4 basically breaks down into three specific lines and things like that Now in a perfect world we might hope that each neuron in the second-to-last layer corresponds with one of these sub components That anytime you feed in an image with say a loop up top like a 9 or an 8 There’s some specific Neuron whose activation is going to be close to one and I don’t mean this specific loop of pixels the hope would be that any Generally loopy pattern towards the top sets off this neuron that way going from the third layer to the last one just requires learning which combination of sub components corresponds to which digits Of course that just kicks the problem down the road Because how would you recognize these sub components or even learn what the right sub components should be and I still haven’t even talked about How one layer influences the next but run with me on this one for a moment recognizing a loop can also break down into subproblems One reasonable way to do this would be to first recognize the various little edges that make it up Similarly a long line like the kind you might see in the digits 1 or 4 or 7 Well that’s really just a long edge or maybe you think of it as a certain pattern of several smaller edges So maybe our hope is that each neuron in the second layer of the network corresponds with the various relevant little edges Maybe when an image like this one comes in it lights up all of the neurons associated with around eight to ten specific little edges which in turn lights up the neurons associated with the upper loop and a long vertical line and Those light up the neuron associated with a nine whether or not This is what our final network actually does is another question, one that I’ll come back to once we see how to train the network But this is a hope that we might have. A sort of goal with the layered structure like this Moreover you can imagine how being able to detect edges and patterns like this would be really useful for other image recognition tasks And even beyond image recognition there are all sorts of intelligent things you might want to do that break down into layers of abstraction Parsing speech for example involves taking raw audio and picking out distinct sounds which combine to make certain syllables Which combine to form words which combine to make up phrases and more abstract thoughts etc But getting back to how any of this actually works picture yourself right now designing How exactly the activations in one layer might determine the activations in the next? The goal is to have some mechanism that could conceivably combine pixels into edges Or edges into patterns or patterns into digits and to zoom in on one very specific example Let’s say the hope is for one particular Neuron in the second layer to pick up on whether or not the image has an edge in this region here The question at hand is what parameters should the network have what dials and knobs should you be able to tweak so that it’s expressive enough to potentially capture this pattern or Any other pixel pattern or the pattern that several edges can make a loop and other such things? Well, what we’ll do is assign a weight to each one of the connections between our neuron and the neurons from the first layer These weights are just numbers then take all those activations from the first layer and compute their weighted sum according to these weights I Find it helpful to think of these weights as being organized into a little grid of their own And I’m going to use green pixels to indicate positive weights and red pixels to indicate negative weights Where the brightness of that pixel is some loose depiction of the weights value? Now if we made the weights associated with almost all of the pixels zero except for some positive weights in this region that we care about then taking the weighted sum of all the pixel values really just amounts to adding up the values of the pixel just in the region that we care about And, if you really want it to pick up on whether there’s an edge here what you might do is have some negative weights associated with the surrounding pixels Then the sum is largest when those middle pixels are bright, but the surrounding pixels are darker When you compute a weighted sum like this you might come out with any number but for this network what we want is for activations to be some value between 0 & 1 so a common thing to do is to pump this weighted sum Into some function that squishes the real number line into the range between 0 & 1 and A common function that does this is called the sigmoid function also known as a logistic curve basically very negative inputs end up close to zero very positive inputs end up close to 1 and it just steadily increases around the input 0 So the activation of the neuron here is basically a measure of how positive the relevant weighted sum is But maybe it’s not that you want the neuron to light up when the weighted sum is bigger than 0 Maybe you only want it to be active when the sum is bigger than say 10 That is you want some bias for it to be inactive what we’ll do then is just add in some other number like negative 10 to this weighted sum Before plugging it through the sigmoid squishification function That additional number is called the bias So the weights tell you what pixel pattern this neuron in the second layer is picking up on and the bias tells you how high the weighted sum needs to be before the neuron starts getting meaningfully active And that is just one neuron Every other neuron in this layer is going to be connected to all 784 pixels neurons from the first layer and each one of those 784 connections has its own weight associated with it also each one has some bias some other number that you add on to the weighted sum before squishing it with the sigmoid and That’s a lot to think about with this hidden layer of 16 neurons that’s a total of 784 times 16 weights along with 16 biases And all of that is just the connections from the first layer to the second the connections between the other layers Also, have a bunch of weights and biases associated with them All said and done this network has almost exactly 13,000 total weights and biases 13,000 knobs and dials that can be tweaked and turned to make this network behave in different ways So when we talk about learning? What that’s referring to is getting the computer to find a valid setting for all of these many many numbers so that it’ll actually solve the problem at hand one thought Experiment that is at once fun and kind of horrifying is to imagine sitting down and setting all of these weights and biases by hand Purposefully tweaking the numbers so that the second layer picks up on edges the third layer picks up on patterns etc I personally find this satisfying rather than just reading the network as a total black box Because when the network doesn’t perform the way you anticipate if you’ve built up a little bit of a relationship with what those weights and biases actually mean you have a starting place for Experimenting with how to change the structure to improve or when the network does work? But not for the reasons you might expect Digging into what the weights and biases are doing is a good way to challenge your assumptions and really expose the full space of possible solutions By the way the actual function here is a little cumbersome to write down. Don’t you think? So let me show you a more notationally compact way that these connections are represented. This is how you’d see it If you choose to read up more about neural networks Organize all of the activations from one layer into a column as a vector Then organize all of the weights as a matrix where each row of that matrix corresponds to the connections between one layer and a particular neuron in the next layer What that means is that taking the weighted sum of the activations in the first layer according to these weights? Corresponds to one of the terms in the matrix vector product of everything we have on the left here By the way so much of machine learning just comes down to having a good grasp of linear algebra So for any of you who want a nice visual understanding for matrices and what matrix vector multiplication means take a look at the series I did on linear algebra especially chapter three Back to our expression instead of talking about adding the bias to each one of these values independently we represent it by Organizing all those biases into a vector and adding the entire vector to the previous matrix vector product Then as a final step I’ll rap a sigmoid around the outside here And what that’s supposed to represent is that you’re going to apply the sigmoid function to each specific component of the resulting vector inside So once you write down this weight matrix and these vectors as their own symbols you can communicate the full transition of activations from one layer to the next in an extremely tight and neat little expression and This makes the relevant code both a lot simpler and a lot faster since many libraries optimize the heck out of matrix multiplication Remember how earlier I said these neurons are simply things that hold numbers Well of course the specific numbers that they hold depends on the image you feed in So it’s actually more accurate to think of each neuron as a function one that takes in the outputs of all the neurons in the previous layer and spits out a number between zero and one Really the entire network is just a function one that takes in 784 numbers as an input and spits out ten numbers as an output It’s an absurdly Complicated function one that involves thirteen thousand parameters in the forms of these weights and biases that pick up on certain patterns and which involves iterating many matrix vector products and the sigmoid squish evocation function But it’s just a function nonetheless and in a way it’s kind of reassuring that it looks complicated I mean if it were any simpler what hope would we have that it could take on the challenge of recognizing digits? And how does it take on that challenge? How does this network learn the appropriate weights and biases just by looking at data? Oh? That’s what I’ll show in the next video, and I’ll also dig a little more into what this particular network we are seeing is really doing Now is the point I suppose I should say subscribe to stay notified about when that video or any new videos come out But realistically most of you don’t actually receive notifications from YouTube, do you ? Maybe more honestly I should say subscribe so that the neural networks that underlie YouTube’s Recommendation algorithm are primed to believe that you want to see content from this channel get recommended to you anyway stay posted for more Thank you very much to everyone supporting these videos on patreon I’ve been a little slow to progress in the probability series this summer But I’m jumping back into it after this project so patrons you can look out for updates there To close things off here I have with me Lisha Li Lee who did her PhD work on the theoretical side of deep learning and who currently works at a venture capital firm called amplify partners Who kindly provided some of the funding for this video so Lisha one thing I think we should quickly bring up is this sigmoid function As I understand it early networks used this to squish the relevant weighted sum into that interval between zero and one You know kind of motivated by this biological analogy of neurons either being inactive or active

(Lisha) – Exactly (3B1B) – But relatively few modern networks actually use sigmoid anymore. That’s kind of old school right ?

(Lisha) – Yeah or rather ReLU seems to be much easier to train

(3B1B) – And ReLU really stands for rectified linear unit (Lisha) – Yes it’s this kind of function where you’re just taking a max of 0 and a where a is given by what you were explaining in the video and what this was sort of motivated from I think was a partially by a biological Analogy with how Neurons would either be activated or not and so if it passes a certain threshold It would be the identity function But if it did not then it would just not be activated so be zero so it’s kind of a simplification Using sigmoids didn’t help training, or it was very difficult to train It’s at some point and people just tried relu and it happened to work Very well for these incredibly Deep neural networks.

(3B1B) – All right Thank You Lisha for background amplify partners in early-stage VC invests in technical founders building the next generation of companies focused on the applications of AI if you or someone that you know has ever thought about starting a company someday Or if you’re working on an early-stage one right now the Amplify folks would love to hear from you they even set up a specific email for this video [email protected] so feel free to reach out to them through that

Treasure

Omg…thats amazing ,well explained sir !

Great video! Very well explained!

Great content but the music in the background is fairly distracting.

Incredibly simple explanation for an initially perplexing topic. Your series has helped piece together readings from multiple textbooks and multiple lectures in my engineering classes at university. Thank you so much for the thought and effort.

awesome video, but they way they teach it to a machine stay a nonsense, neural network is great but I would never train my machine to learn by this way, my network would be more complicated maybe 10 layout but it could solve entire vision not only numbers, but the video is cool and I will never be able to code

algorithm is not coding

After 5 years on a technical university and several courses on machine learning, I think this is the best explanation I have ever seen. Bravo!

“By using the Japanese K supercomputer, with 700,000 processor cores and 1.4 million gigabytes of RAM, a whopping 1.7 billion neurons and 10 trillion synapses were simulated. It took the darn thing 40 minutes to simulate 1 second of ‘activity’”

– Forbes

Scroll back up and learn this!

Sigmoid Squishification.

– 3B1B, 2017.

You can technically teach kids how to program deep learning easily.

Fist key is under standing like why is the sky blue

thank you very, very much. i finally understand how the weights and biases work. years ago, i tried to program neural nets. but i just could not comprehend the weights and biases. you explain them in such a simple way that i am going try again to program neural nets. what programming language is the best to implement neural nets these days?

动画制作的非常优秀，非常简洁明了的吧原理展现给我们了，谢谢up主

Wreckit Ralph will smash it!

lost it after min 8….

Isn't there a mistake at 14:40? Shouldn't the vector of biases be indexed until k not n ?

Since these neurons each have values between 0 and 1, i guess neural Networks can be made really complex using Quantum computers. Qbits are also not operating static between 1‘s and 0‘s so basically instead of using this on a programmed level, which needs code, this whole thing should be able to be set up on a processor level and be much more efficient.

下次一定

I enjoyed your video and the small roast of the company at the end

Isn't the Bias vector supposed to be b0,…,bk instead of b0,…,bn?

"a couple OF" things, not "a couple" things.

"sigmoid squishification". . yeah.

5:48 hacking! MWAHAHAHA

I came here after deepfake.

Great video! My course this semester is based on Bertsekas's "Reinforced learning and optimal control (draft version)". Due to it still being a draft I imagine, it's extremely generic. For example it introduces neural networks with the compact formula and no examples or analytical explanation. Thanks to this video (and this video alone) I now have some genuine understanding in what is going on.

Ali Yahya seviliyorsun kardeşim.

I think I couldn't find more clear video than this one, thanks a lot for this masterpiece!

VERY easy to follow explanation, thanks

このビデオ、グラフィカルに説明していて、数学的な部分も非常にわかりやすいですね。

ところで、日本人いませんかね？

I am technically a neural network

Dude, never expected such an explanation!! Wooow like i don't know how to thank you!! Really great job!

This is pretty much the same mistakes that the suggested word for the same was mistakes to be honest about the ur in a company that has been used for years now because bean was translated by another fierce threat short-term Beat from the start hex editing is that the current contract fact that thus is the most accurate part of the story nowadays apparently there has never various Scare tactics that mattered.. cutting edge of the situation as soon as possible for you to realize that canhave was the best thing to happen really WITH being used as a person I have thought about how much I have to do to help myself and make sure ICU is okay I'm just glad I can not extend the call UC it makes me to do that but hell yeah this time it was a very long time as a joke.. To serious decisions about my unimportant life yet ahead of time to the rest of my own life because of something happened to ME Worser than the other js and then Linux to the Words of God that has never been a thing in my entire English language I was homeless and living in a very REAL LIFE RIGHT NOW HAS BEEN THE SOURCE CODE TO THE WAY OF 1ST FOR THE FIRST TUNED ERA IS HIP TO BE SQUARE HI on Facebook with great judgement in critical reception stuff like this year is a very unusual one and historically that much of the application of the myriad of the artificial intelligence advances and then the machine learning algorithms &should make a lot to do with the fact that it's dangerous life right now in my opinion go to a different path FYI him do not have any other side concerns done this time how many days your own unique way to respond no to the other two really is the mystery that boggles the Mind to see what happens to the two people suggested to be a problem for some reason and I would never consider it a Priority to Consider that may have been a way to help me understand that I am with great understanding now since the way memories for me seem to flash back until another

Please never stop making such videos… Your videos have helped me a lot. Thanks a ton. From India

What you say to yourself your Subconscious gives birth to! You are consciousness and you and your Subconscious are two parts of the same One that you are.

amazing talk!!

uma obra de arte esse video!

A sincere question (hoping someone could answer): what's the problem with a method with zero hidden layers? That is, simply optimize to 784 * 10 parameters that apply to 784 numbers and return a value between 0 and 1, with 0-0.1 indicating the number '0', 0.1 – 0.2 indicating the number '1', etc.

…awesome

The animation is impressive.

The real question is what if you have been living under a rock😐

Great video, learned a lot! One question: doesn't the location parameters of each neuron have be entered into the input layer for the activations to be meaningful?

Great channel!

This is genius… First time i have watched such a good video that explain this complex topic with this much ease..subbed

One of the few teachers that don't make you feel stupid, but actually help you understand the topic. I appreciate the time you spend on this.

Just found it. It is a great video!!!!!!! Thanks!! I ganna watch all the videos here!! Thanks for creating it and charing

Richest 1% may own 50% of all money, but

3Blue1Brown owns 55% of humanity's total IQs!

now i feel like i am a robot not human anymore

But can the ReLU output be greater than 1? How does this work?

Your brain has no idea how it works

Amazing channel n_n

So weights in neural network work a bit like probability, right ? The more it has weight the more it is probable it has a certain pattern. Am I correct ?

I am honestly mind blown on how something so complicated can be explained so simply …….. This guy is a neural net that takes complicated concepts like deep learning and outputs it in words us mortals can understand. Thank you sir.

Dumb computers

I am gonna name my dog squishifoo.

Full AGI release!:

https://aidreams.co.uk/forum/index.php?topic=14490.0

HOW R ALL UR VIDEOS SO GOOD AND INTUITIVE MY MAN?

The more you know, the more you know that you don't know.

This video is awesome! It's extremely well explained, so much that I think I could try to code this my self in Python just thanks to the explanation!

ohh yeah the EDGE. KNOBS and

I actually subscribed to the channel because I want Youtube's recommendation algorithm to give me more like this

Please make video on LSTM and CNN as well.

No no watts teachings, no neural to technology…teachings

11:35 보는중 (자막x)

Really nicely explained.

Instructions unclear. I now have dyscalculia.

<3

I watched this video when it came out and understood like a third of it. 2 years of math classes later, and now I understand the first 14 minutes of it. Then he hit me with that linear algebra.

this was shockingly similar to a projection matrix. didn't see that coming

简单明了

Thanks a lot. Excellent video. Inspiring me for a proposal for my research director. Greetings from Popayan, Colombia.

Great job !

Il piano astrale e' più che reale e' del tutto artificiale connesso alla matrix universale

I wanted to learn about machine learning before, I’m really interested in AI. This is really helpful, thanks!

This man makes calculus seem like prealgebra

Why I need that waited sum as a math framework of activation functions or for determining activation value?

I can´t belief how good and easy this video is! Well done, very helpful

Great content!

Dude you are probably the most impressive teacher I have ever seen…congratulations you are a blessing

8:50 Here's where I lost the track and got confused

Very beautifully explained neural networks !

Thank you for all these beautiful videos 🙌

If only he had better morals, he might actually accomplish something for humanity…….

greet~~~

at 00:56, you probably said it wrong, it should be 0 to 9, not 0 to 10. Anyway that does change the concept though 😉

I think there was a mistake at 14:42. The matrix for the bias should go from b0 to bk, not bn.

Awesome video regardless.

Can you tell me about the hidden layer and the number of hidden layer depend upon??

Hi, I have a doubt. what is the difference between parameter a and w ?

Typo: At 14:40 the vector of biases that is added should be [b0…bk] and not [b0…bn] because the size of the matrix that you get from the multiplication is [k+1] x 1 and not [n+1] x 1.

Very informative and interesting.

At 10:05 you say that adding the negative weights around to detect edges will increase the weighted sum, but surely as the activation value is between 0 and 1, it will decrease the weighted sum and thus affect which neurons are affected in the next layer?

There is an error at the matrix indexes at 16:00, the last column index at first row is n and at other rows it's k, should be consistent

Can’t believe they got Peter Gregory to narrate.

Can anyone help me understand the 9:50 explanation on weights depicting pixels?

f that's the best subscribe request I ever seen at the end of a video: subscribe so the AI can take positive data, on a ai video, noice.

holy crap thank you so much for this video it helped so much

Random thought, imagine watching this when you have trypophobia

you are a God Gifted Teacher! please accept my respect master!!!

So inside one neuron, do each weight for each neuron have different values? Or the same weight value for each input in one neuron?

@3blue1Brown At 14:38 isn't matrix for bias supposed to be [k x 1]? instead of [n x 1]? I am not sure if I'm right but I think since there are k neurons at layer 1, the number of bias also should be k?