How to Make a Neural Network - Intro to Deep Learning #2

How do we learn? In this video, I'll discuss our brain's biological neural network, then we'll talk about how an artificial neural network works. We'll create our own single layer feedforward network in Python, demo it, and analyze the implications of our results. This is the 2nd weekly video in my intro to deep learning series (Udacity nanodegree)

The coding challenge for this video:

Ludo's winning code:

Amanullah's runner up code:

Please subscribe! And like. And comment. That's what keeps me going.

More learning resources:

Please support me on Patreon:
Follow me:
Signup for my newsletter for exciting updates in the field of AI:

[nsudhanva]

You gotta appreciate the effort he puts into his videos! Those raps and memes! Amazing video. Thanks!

[RenjiB]

LOL @ "biochemical warfare" 1:38

[alishaaneja]

Your videos are perfect for me! I am someone who will sleep during a lecture or talk, unless the speaker is very enthusiastic and engaging.
You are putting a lot of effort in these videos and it’s commendable!

[LouisFENDJI]

I am not used commenting videos, but this time I cannot resist to do it. YOU ARE THE BEST! I send the link to one of my student who is learning machine learning right now. Thumbs up!

[ClaudiaOfTheWorld]

siraj ur a boss, only youtuber i don't have to put on 1.5x speed! never slow down ! fire those action potentials !

[MyHandsAreStuck]

I'm so confused. I wish you went slower and explained more.

[guillempitarch]

6:44
Aaaaand you just gave me my new wake-up motivational tone, thank you master

[kshitizk]

All this makes sense to me in one go only after watching few weeks videos of Machine Learning - Andrew Ng!

[mayanksaxena]

Congratulations for 50k subscribers!

Loving the rap <3

[geoglyphiks]

Kali Hubud, neurophysiology, climate change, python overview with meme overlays for entertainment value? Dude. You're speaking my language. Really great video, appreciate all the artistry and effort you put into this. Subscribed.

[JSD]

i cant create such wonderfull effort with Raps to memorise it easily ..Great effort Siraj

[mlfan]

thank you for sharing and teaching. you are the next generation of education!!
ML for the mortals!!

[dnisgd]

What do you think would be great ressources to start with if you wanted to get more deeply into machine learning? Does it make sense to try a bottop up approach, where you try to look at what the machine is actually doing on an atomic scale, and then see how an operating system and software runs on it? Or does it make more sense to simply try and understand the python basics and kind of "accept" that a computer is able to do certain things? If you wanted to get actual understanding of what's happening, what would in your opinion be the best way to go at it?

[xisco]

First of all, thanks for all the videos you upload, the topics you cover are a great source for DL/ML implementations with python and to get to know AI theory from a more practical perspective. I totally respect the way you communicate, but in my opinion, when watching the videos I feel like I am watching a James Carrey motivation video, a BMW video ad, all mixed up with a great explanation of the topics. This is just a personal opinion. So far, from a technical point of view the content is great. Thanks.

[DubstepDinosaurs]

I love the cameo girl! Your "67 questions" vid with her was great. I still think of her comment about how in 2017, machines will solve problems that benefit our lives, and programmers will have little idea how they are doing it, because of deep learning.

[Xaminn]

Your videos are quite funny and very educational. Thanks for the great videos!

[saravanabalagi]

Feedback: (Hey @SirajRaval, you did a good job there for coding ANN from scratch. This is not criticism but ways in which your teaching can be improved so it reaches the learners deep into the brain)

I had numerous doubts that popped spontaneously as I listened to this video, some of which are:
1) Why do we use sigmoid function?
2) Why do we use gradient descent?
3) Why are we iterating over a thousand times?

I would like the same to be taught in this way: The objective is to find the weights in such a way that error (sum of squares of individual error) is minimum, and we shall do it in any way we want to. But since those other ways incl. brute force is tedious (explaining it with a time lapse video) we finally use gradient descent. Similarly, we could have used tanh or something else there (explaining the problems we will face if used), but here's why we use sigmoid... [I'm no expert in ANN I just learnt it just now from Welch Labs demystifying NN videos]

[PanetMaster]

Love your work! I just finished Andrew Ng's intro to machine learning course on Coursera and now I'm starting to code along with your videos. It's a great way to keep learning. Just a note, it looks like you renamed the "predict" function "think", resulting in the code not working right out of the box.

[hikingnerd]

Amazing! These are the best machine learning videos.

[cristinaheghedus]

Love your videos! You wanted 50.000 subscribers and you have it! Just keep on going, you have the attitude to reach everything you want! Congrats :D I myself am interested in Machine Learning, Deep Learning... and cannot wait to open your video when you upload it!