Neural Networks from Scratch – P.3 The Dot Product
[ad_1]
Neural Networks from Scratch book: https://nnfs.io
NNFSiX Github: https://github.com/Sentdex/NNfSiX
Playlist for this series: https://www.youtube.com/playlist?list=PLQVvvaa0QuDcjD5BAw2DxE6OF2tius3V3
Neural Networks IN Scratch (the programming language): https://youtu.be/eJ1HdTZAcn4
Python 3 basics: https://pythonprogramming.net/introduction-learn-python-3-tutorials/
Intermediate Python (w/ OOP): https://pythonprogramming.net/introduction-intermediate-python-tutorial/
Mug link for fellow mug aficionados: https://amzn.to/3cKEokU
Channel membership: https://www.youtube.com/channel/UCfzlCWGWYyIQ0aLC5w48gBQ/join
Discord: https://discord.gg/sentdex
Support the content: https://pythonprogramming.net/support-donate/
Twitter: https://twitter.com/sentdex
Instagram: https://instagram.com/sentdex
Facebook: https://www.facebook.com/pythonprogramming.net/
Twitch: https://www.twitch.tv/sentdex
#nnfs #python #neuralnetworks
Source
[ad_2]
If you Python-Fetishists out there want a one-liner here you go: output = [sum([i * w for i, w in zip(inputs, weights)]) + bias for weights, bias in zip(weights_list, bias_list)] 🙂
animations are really helpful
THE ANIMATION IS MIND BLOWING!!!
COMING FROM ENGINEERING BACKGROUND MYSELF, I found very impressive how you use the straight line equation and visualiSation to make understand the meaning of BIAS AND WEIGHT, THANK YOU LEGEND SENTDEX
THE WAY you EXPLAIN IS FOR DUMMIES,fantastic!!! THE JAPANESE GUY CS DOJO is also good like you at explaining using animation
Who needs college degrees when we have sentdex
I want to build a neural network that can predict which one of his coffee mug collection he will use for the next episode ☕. Brilliant series btw, easy to understand and to the point. 👍
what text editor is this??
I love this series. I wanted to learn everything from scratch to know how actually deep learning and machine learning works.
Luckily I found you on youtube. So when are you uploading Part 6? I'm waiting for it.
I'm trying to implement further things on my own like cost function, optimizer, backpropagation
But please upload more videos as soon as possible.
Thanks for this amazing series.
Does (4, ) mean the same as (4,1) ?
Looks like np.dot(inputs, weights) here will use the number of rows of the matrix (3), if you drop an input and add a bias, and then dot the np.dot(inputs, weights) you won't have any shape error messgae, but obv u wont have same result.. Still unclear why the order matters tho
I'm getting the nnfs book today as an early Christmas present to myself!
You can't have a neural network with biases only, because the output of the neuron would always only be whatever the bias is, and any change in the input would not change the output. In fact, you'd have to completely rely on backpropagation to do the job of the weights for you, so you'd essentially need to re-train the network for every single observation, and then it can only model that observation.
Imagine trying to model any curved surface using only horizontal lines that you can shift up or down.
OH MY GOD I LOVE YOU! The explanation you gave for them 3D arrays was the AHAAAA!! moment for me.
Man, the animations are in my opinion fundamental for the full understanding of the content. Huge thanks to Daniel who's done them.
your channel is more exciting now.
Wow! Nice video, very easy to understand for beginners like me. Shout out sa animator, very helpful nga talaga, lalo na sa part ng "Shape". Thanks!
I really do not understand how you end up with 2.8 by doing some sort of an operation on two lists [1.0, 2.0, 3.0, 2.5] and [0.2, 0.8, -0.5, 1.0] ? What am I missing…
these videos are just like drug, i can't stop watching and I want more and more
"list of lists, lol for short" i have another word in my dictionary now
Can you do it with recursion?
I recommend the https://colab.research.google.com/ for the ones who doesn't want to have problems with installing python and dealing with virtual environment. It's an online jupyter notebook platform from google, where you can run it locally or on google servers. It's pretty neat and I use it a lot
You explain in such a nice and easily understandable way and plus those animations are very helpful.
It's so helpful content and because of you only I'll pass my college exams
Thanks
So weight is like the slobe, and bias is like the y intercept?
Nobody:
Programmers: LOLOLOLOLOL
Really excellent series!! Nice work.
Whats the intuition behind using a dot product as opposed to some other way to combine weights and outputs?
The animations are indeed fantastic, great job!
imo in this case at 17:52 is beacuse of basic matrix operation principle
a = (1, 2, 3) -> 3 colomn in 1 row
b = (1, 2) -> 2 colomn in 1 row
you have to invert it, like a*b is 3 colomn in 1 row * 1 colomn in 2 row
meaning you have to arrange the rows and colomns is an acceptable way, [4 col, 1 row] cannot be multiplied with [4 col, 3 row]
if i remember correcly
Sentdex update your account to receive tips from Brave users. I just sent one.
how do i install numpy?
3b1b's linear algebra I believe is mandatory watching for anything that has it as a prerequisite
can we have part 6 please?
Amazing
how about a visual studio version? 😛
What if I tell you complex space is a tensor product 😀
You are a maestro!
Greatings from Italy
Finally somebody is starting explaining how NN work so anybody can understand and start building there own Ai thank you very much! You are a good teacher!
how do I get numpy with PyCharm.
dude, your video is very helpful. i wish to say thanks for uploading these
amazing video , i've learned alot from it , but is that a burger mug 🙂
At first I was like wow that's a huge burger !!!
Why would he do a tutorial while eating a burger.
Then I realized I was tricked and it was actually a mug.
Hey sentdex, why does the addition of the bias to the dot product actually add the numbers instead of appending the list like it would happen if you try to add two lists together?
Love those animations, I was taking course from coursera too, but it was so confusing! you made this much clear to me!
Doesnt the dot product need one of the matrices to be transposed ? Isnt it the basic rule…if it doesnt apply here.. please tell y?…. I think the best way to do dot product or to teach dot product would be using transpose…… np.dot(inputs.T, weights)…please correct me if i am wrong
9:07 "In mathematics, a list is a vector". I thought in mathematics, a vector was a measurement with magnitude and direction. Can someone please clarify this for me? Thanks.
great video
really easy to follow
the animations really helps to visualize the concepts you are trying to explain
i think if you guys make the brackets in pairs with the same color it's going to be easier to see (just for the part like the lololo)
[[]] that's how i'm laughing online from now on
how can anyone vote this down? good vid series thx