Neural Networks: Crash Course Statistics #41

Today we're going to talk big picture about what Neural Networks are and how they work. Neural Networks, which are computer models that act like neurons in the human brain, are really popular right now - they're being used in everything from self-driving cars and Snapchat filters to even creating original art! As data gets bigger and bigger neural networks will likely play an increasingly important role in helping us make sense of all that data.

Special thanks to Max Deutsch giving us permission to use his Neural Network-written Harry Potter chapter in this episode! Read more about his work here:

Neural Network to create anime characters:

Neural Network to great Van Gogh-like art:

Neural Network to create skate decks:

Thanks to the following Patrons for their generous monthly contributions that help keep Crash Course free for everyone forever:

Sam Buck, Mark Brouwer, Jennifer French Lee, Brandon Westmoreland, dorsey, Indika Siriwardena, James Hughes, Kenneth F Penttinen, Trevin Beattie, Satya Ridhima Parvathaneni, Erika & Alexa Saur, Glenn Elliott, Justin Zingsheim, Jessica Wode, Eric Prestemon, Kathrin Benoit, Tom Trval, Jason Saslow, Nathan Taylor, Brian Thomas Gossett, Khaled El Shalakany, SR Foxley, Sam Ferguson, Yasenia Cruz, Eric Koslow, Caleb Weeks, Tim Curwick, D.A. Noe, Shawn Arnold, Malcolm Callis, Advait Shinde, William McGraw, Andrei Krishkevich, Rachel Bright, Kathy & Tim Philip, Jirat, Ian Dundore
--

Want to find Crash Course elsewhere on the internet?

[Antenox]

Those reCAPTCHAs are themselves part of a neural network. They're designed so that users help teach image recognition AIs, originally for text, but now for street images.

[oliverschmid]

Crazy to see how much text generation has changed in four years. From nonsense to chatgpt

[mananmaheshwari]

Amazing how u covered so much of deep learning within 12 minutes. Very clear and informative.

[teegnas]

Probably the best layman's guide to neural network

[scottjoewill]

I've seen and read a lot of explanations on how Neural Networks...work, but this is by far the best and easiest to understand one yet. Thanks for making learning easy. Also, shout out from Indy! Woo!

[SunriseFireberry]

I've looked at data from both sides now
...I really don't know data
at all

[pacatrue]

8:16. I'm going with John Green is #1, Stan is #2 and Hank is #3.

[gooseontheinternet]

7:03 "so, jk rowling isn't out of a job **yet**."

[grainfrizz]

If you're someone interested in really learning Machine Learning (specifically multilayer perceptron/artificial neural networks), go watch this to get a high level understanding. And then afterwards, watch 3Blue1Brown's video about it. And then afterwards, get your hands dirty by watching (and practicing as well) The Coding Train (Daniel Shiffman) playlist about build his own Neural Network JavaScript library.

Avoid Siraj Raval's video as they're not helpful—only cool to watch.

[PestOnYT]

We humans always have an option we deny those NNs. The "I have no clue" option. At 2:10 in the video we force it to make a judgement without offering an escape route. A common flaw in today's NN.

[ziggyoickle]

11:17 I think there was an audio problem, it only worked for me when I unplugged my headphones past this point, I'd suggest unplugging for the smoothest experience as soon as you see this message

[jordanmakesmaps]

Lots of explanations of deep learning on Youtube, but this is a really good one!

[reinforcer]

currently working with lstm to for forecasting prediction and autoencoders for deep representational learning of patients' MRI volume data!

[ianrbuck]

Oi, don't intentionally trigger all my voice-activated assistants!

[zertilus]

I had a weird idea for crash course videos and educational videos in general. Instead of a speaker reading lines to a viewer, you make a version where the speaker is reading lines to a sit in, like Hank or Micheal. All the listener would do is mhm and okay things, or more importantly repeat and reword the lines. Better reinforcement, and more natural in the way people are going to apply the knowledge.

[informativecontent]

this is by far one of the best channels for education and learning skills thanks for being there

[amabitsapiens]

Incidentally, the example neural network (albeit cursory for the sake of a quick theoretical construct) is critically missing "health" and "# of dependents". A recent study discovered that mothers consistently are paid less for the same job title than non-mothers.

[Felishamois]

Crash Course General Western Music Theory (with Jacob Collier?)
Crash Course "Indian Music Theory" (??) (a person who knows their stuff, not a lot of resources out there for english-speakers)
Crash Course Music Theories (with Herbie Hancock?)
Crash Course Written (worldwide, like really) Music History (with Adam Neely? David Hudry?)
Crash Course Popular (worldwide, like really) Music History (with Rick Beato?? Hank Green?)
this could happen. in the next ten years. it's doable. keep up the great work.

[ganaraminukshuk]

You had me at neural networks but then I remembered that NN's require a lot of prerequisites to fully understand.

[dwrhpdu]

I turned on my p c first it took a few minutes to update windows, and than, i think before i klicked on the mozzila icon, it threw me this video for no apparent reason... Honestly