But what is a Neural Network? | Deep learning, chapter 1


[ad_1]

Home web page: https://www.3blue1brown.com/
Brought to you by you: http://3b1b.co/nn1-thanks
Additional funding offered by Amplify Partners

Full playlist: http://3b1b.co/neural-networks

Typo correction: At 14 minutes 45 seconds, the final index on the bias vector is n, when it is purported to in actual fact be a ok. Thanks for the sharp eyes that caught that!

For those that wish to be taught extra, I extremely suggest the e book by Michael Nielsen introducing neural networks and deep studying: https://goo.gl/Zmczdy

There are two neat issues about this e book. First, it is obtainable totally free, so take into account becoming a member of me in making a donation Nielsen’s method when you get one thing out of it. And second, it is centered round strolling by some code and information which you’ll be able to obtain your self, and which covers the identical instance that I introduce on this video. Yay for energetic studying!
https://github.com/mnielsen/neural-networks-and-deep-learning

I additionally extremely suggest Chris Olah’s weblog: http://colah.github.io/

For extra movies, Welch Labs additionally has some nice sequence on machine learning:
https://youtu.be/i8D90DkCLhI
https://youtu.be/bxe2T-V8XRs

For these of you seeking to go *even* deeper, take a look at the textual content “Deep Learning” by Goodfellow, Bengio, and Courville.

Also, the publication Distill is simply completely lovely: https://distill.pub/

Lion picture by Kevin Pluck

—————–
Timeline:
0:00 – Introduction instance
1:07 – Series preview
2:42 – What are neurons?
3:35 – Introducing layers
5:31 – Why layers?
8:38 – Edge detection instance
11:34 – Counting weights and biases
12:30 – How studying relates
13:26 – Notation and linear algebra
15:17 – Recap
16:27 – Some closing phrases
17:03 – ReLU vs Sigmoid

——————
Animations largely made utilizing manim, a scrappy open supply python library. https://github.com/3b1b/manim

If you wish to test it out, I really feel compelled to warn you that it is not probably the most well-documented software, and has many different quirks you may anticipate in a library somebody wrote with solely their very own use in thoughts.

Music by Vincent Rubinetti.
Download the music on Bandcamp:
https://vincerubinetti.bandcamp.com/album/the-music-of-3blue1brown

Stream the music on Spotify:
https://open.spotify.com/album/1dVyjwS8FBqXhRunaG5W5u

If you wish to contribute translated subtitles or to assist assessment people who have already been made by others and wish approval, you possibly can click on the gear icon within the video and go to subtitles/cc, then “add subtitles/cc”. I actually recognize those that do that, because it helps make the teachings accessible to extra folks.
——————

3blue1brown is a channel about animating math, in all senses of the phrase animate. And you already know the drill with YouTube, if you wish to keep posted on new movies, subscribe, and click on the bell to obtain notifications (when you’re into that).

If you’re new to this channel and wish to see extra, a good place to start out is this playlist: http://3b1b.co/advisable

Various social media stuffs:
Website: https://www.3blue1brown.com
Twitter: https://twitter.com/3Blue1Brown
Patreon: https://patreon.com/3blue1brown
Facebook: https://www.fb.com/3blue1brown
Reddit: https://www.reddit.com/r/3Blue1Brown

supply
[ad_2]

Comment List

  • 3Blue1Brown
    November 10, 2020

    Thanks for explaining the concept crystal clear!

  • 3Blue1Brown
    November 10, 2020

    Oh i get the middle layers thing.

  • 3Blue1Brown
    November 10, 2020

    There’s a more symmetric version of ReLU
    Max(-1,x)-Max(1,x)+1 is the thing

  • 3Blue1Brown
    November 10, 2020

    I am not a robot

  • 3Blue1Brown
    November 10, 2020

    there's god sitting in between those patterns.

  • 3Blue1Brown
    November 10, 2020

    Sounds a little like Sudoku.

  • 3Blue1Brown
    November 10, 2020

    No words for ur efforts sir.
    U just make me enjoy stuff …. thanks a lot

  • 3Blue1Brown
    November 10, 2020

    One day ill be able to solve a captcha

  • 3Blue1Brown
    November 10, 2020

    Thank you so much for the visualizations! I think if math teachers always refer to how it pictorially looks the intuition would be easier to get. This is why you and Dr. Strang from MIT have good content!

  • 3Blue1Brown
    November 10, 2020

    Brainchip.

  • 3Blue1Brown
    November 10, 2020

    Here, have my sub

  • 3Blue1Brown
    November 10, 2020

    Do this in c programming

  • 3Blue1Brown
    November 10, 2020

    How can I learn it good?

  • 3Blue1Brown
    November 10, 2020

    Please answer my question:
    Who decides what pattern which neuron will try to detect?

  • 3Blue1Brown
    November 10, 2020

    Let's do a complete overhaul of cultures.

  • 3Blue1Brown
    November 10, 2020

    This guy: Uses dark screen to illustrate a long concept so that it's easy on the eyes.
    Everyone: "Carefully, he's a hero."

  • 3Blue1Brown
    November 10, 2020

    Amazingly explained! thanks so much.

  • 3Blue1Brown
    November 10, 2020

    ehh third attempt lets start

  • 3Blue1Brown
    November 10, 2020

    Hands down, one of the best channels on youtube!

  • 3Blue1Brown
    November 10, 2020

    so many thumbs down? WTF???

  • 3Blue1Brown
    November 10, 2020

    Moral of the story: Don't let your AI watch this, probably they can solve r captcha problems

  • 3Blue1Brown
    November 10, 2020

    SENTDEX RECOMMENDED ME THIS YT CHANNEL

  • 3Blue1Brown
    November 10, 2020

    I'm missing the music we're used to at the beginning 😔

  • 3Blue1Brown
    November 10, 2020

    My first time seeing any of your videos
    And I have to say I'm IMPRESSED. subbed

  • 3Blue1Brown
    November 10, 2020

    Hey! Does that mean the game Battleships is basically a primitive fourm of code??

  • 3Blue1Brown
    November 10, 2020

    brain makes efforts for that. you are wrog. brain spends energy for that. brain spends energy at least %20 of which total body's produced energy for working..

  • 3Blue1Brown
    November 10, 2020

    Excellent video. Please check the Course Mentor on the link below as well.

  • 3Blue1Brown
    November 10, 2020

    Wait i am 10 and trying to understand

  • 3Blue1Brown
    November 10, 2020

    Absolutely phenomenal video. Really shocked to see information like this be explained so succinctly and clearly. This answered a lot of questions regarding how these networks function. Thank you.

  • 3Blue1Brown
    November 10, 2020

    slow down bro, my main language is not English, you use too many new words and talk so fast, so please help peoples like me

  • 3Blue1Brown
    November 10, 2020

    I am having the feeling that in few years most of the machines are going to be more inteligent than me, but if that make me my live easier, is ok !

  • 3Blue1Brown
    November 10, 2020

    Wow. I did my PhD in what was basically image processing and machine decision making (with a bit of machine learning in there), am about as comfortable with linear algebra as any engineer gets, and STILL could not find a "quick" explanation of neural networks that actually taught me something. Everything was either entirely too fundamental–basically explaining how cross-correlation can help identify something using one or another Cauchy-Schwarz corollary–or jumped straight into using abstract algebra terms for the system. As an engineer, I want the mathematical meat, but I think in linear algebra terms. Calling a function a "map" before I see the matrix equation will lose me.

    THANK YOU for filling this weird gap in my understanding! I can imagine hand-coding a neural network, now.

  • 3Blue1Brown
    November 10, 2020

    awesome!

  • 3Blue1Brown
    November 10, 2020

    Are you hard coding the fact that you want one of the lines or groups of pixels to be in the top half? Because what if it's a tiny number in the bottom right.

    I hope you aren't hard coding this but you made it sound like you were when you were explaining

  • 3Blue1Brown
    November 10, 2020

    Amazing, sorry for mi english, the IA in the last exercice know I am lost in the last point and surrender.

  • 3Blue1Brown
    November 10, 2020

    This video is seriously gold!

  • 3Blue1Brown
    November 10, 2020

    Holy shit, what an explanation! 10/10

  • 3Blue1Brown
    November 10, 2020

    Keewl

  • 3Blue1Brown
    November 10, 2020

    this neural network should be called captcha killer

  • 3Blue1Brown
    November 10, 2020

    Can this be done in batch

  • 3Blue1Brown
    November 10, 2020

    मस्त… खुप छान…

  • 3Blue1Brown
    November 10, 2020

    3Blue1Brown
    "Sigmoid Squishification Function": 11:23
    Most brilliantly named function I have ever heard named. Absolutely brilliant. The merger of the technical with the simple with a double alliteration for easy memory.

  • 3Blue1Brown
    November 10, 2020

    "A neural network is a lookup table." – Anonymous smart person on the internet.

    I forgot who on the internet told me this, but the person was absolutely right. A neural network is a lookup table that was magically created by a learning process that we can't understand easily… but it's still just a lookup table. And that is the problem with them… they have more inputs than are necessary to get a useful output… and THAT is why they fail when we feed them things they have never been taught. Plain neural networks are entirely incapable of sanity checks.

  • 3Blue1Brown
    November 10, 2020

    every time you feel useless in life remember that you can recognize a number instantly while a super-powerful machine needs hundreds of thousands of examples to do the same

  • 3Blue1Brown
    November 10, 2020

Write a comment