I have enough trouble getting my vectors to multiply correctly in python, I can't imagine working with this all in Js just to port it TO DESMOS. Props to you, and also your desmos 3D rendering videos are sick as h*ll
Hey John! Love your content, and your channel is really what inspires me to make my videos. Your stuff is always so much more impressive and better than mine! And thanks for the plug in the description!
Whoa, one of the most concise explanations of neural networks and fits in an extremely creative idea to a somewhat stale and overused dataset! John is not big brained. He is INTERGALACTIC sized brain!
Wow, 140! subscribers. I didn't know there even were 1.346201e+241 people on earth. All jokes aside, you deserve way more subscribers than you currently have. Keep up the good work
Backpropagation works by taking the error value which is the difference between the expected output and the actual output and multiplying the error by the slope of your activation function which makes the output delta which you can multiply by the input and add to the weight. The hidden delta is a little bit different by taking the output delta multiplying it by the synapse connecting that hidden node to the node ahead of it and then multiplying by the slope of the derivative to get the delta. Delta is applied to all weights similarly and you can repeat the process propagating through the network.
Isn't Backpropagation just Setting right values in those parameters (Mathematically translated Biological functions) Lol why make it so complex to understand?
@@yashaswikulshreshtha1588 Alas you cant just plot the inputs to the outputs and algebraically derive the weights. The only way to find the weights is to slide the weights into place... Or just use a python library. That's a lot easier.
I am gonna start digital protest for not giving this channel enough fame and recognition. You're kind of me because I like to do mind boggling, mind shattering stuff and seeing you not on top because of youtube I am sad.
How do you find all this info about Desmos? I didn't even know you could extract the x and y values of a point until I saw it in someone's Desmos file. Their documentation only adds a little onto what's obvious.
i like how you just pop in every month or two and just say "oh yeah by the way i found ANOTHER way to use desmos in an unintended way"
I've been waiting for you to pop in for seven months buddy
@@JohnDoesStuffLol I am gonna sue RUclips and you for not having more than million subs.. I loveee your videoo
How the heck did you ever think of this? I am genuinely impressed at the creativity more than anything else. Subbed
okay so can you please teach us how to make a line thank you
Absolutely insane that I see a comment posted 2 hours ago from a channel I recognize on a video about Desmos neural networks posted 10 months ago.
@@DanielZubenko01 lmao, welcome to the RUclips algorithm
y = mx + b
m = the slope
b = the y intercept
@@inkeys6244 what are y and x
@@jimblesjoestar8741 ask your mom
I have enough trouble getting my vectors to multiply correctly in python, I can't imagine working with this all in Js just to port it TO DESMOS. Props to you, and also your desmos 3D rendering videos are sick as h*ll
To be fair, Desmos is written in JS, so "port" is a strong word. But it's still impressive.
@@MCLooyverse that's nice at least
Me using DESMOS: Huh, that's a line that moves
This guy using DESMOS: ANYWAYS, THIS IS AN AI THAT RECOGNISES HAND-WRITTEN NUMBERS
With this music, this guy sounds like an evil scientist who found another way to exploit Desmos in unintended ways and is about to conquer the world.
I love the presentation style pace, talk about beaming info straight to our heads
Hey John! Love your content, and your channel is really what inspires me to make my videos. Your stuff is always so much more impressive and better than mine! And thanks for the plug in the description!
Result looks fantastic! Great work.
Whoa, one of the most concise explanations of neural networks and fits in an extremely creative idea to a somewhat stale and overused dataset!
John is not big brained. He is INTERGALACTIC sized brain!
I love the idea of using things to do stuff it was never meant to do. This is amazing!
Wow, 140! subscribers. I didn't know there even were 1.346201e+241 people on earth. All jokes aside, you deserve way more subscribers than you currently have. Keep up the good work
this joke is just... so bad and so good
@Mobcrafter
That's a perfect description lmao
Who’s “You Currently”?
this is like the third video i've watched and dude im so hooked!!!!!! i hope you start uploading again
new sub gang!!!
look who it is
Lmao what, I just finished watching your stopwatch video and now you're here?? Like wth
I subbed, man. A really cool way to present stuff
Magic stuff once again, John. Always dazzled by your vids.
I absolutely love your content! It teaches me so much every time, keep it up!
I'm not big brain enough to understand your videos completely but I still enjoy watching them. Your effort in these videos got you a sub
the second you explained that chunk of math my world just opened up. RESPECT.
Just discovered your channel, this is pure gold.
I absolutely love it when I find a small channel that make a high quality content!
you thought we wouldnt hear the demonic little gray cat theme in the background
Backpropagation works by taking the error value which is the difference between the expected output and the actual output and multiplying the error by the slope of your activation function which makes the output delta which you can multiply by the input and add to the weight. The hidden delta is a little bit different by taking the output delta multiplying it by the synapse connecting that hidden node to the node ahead of it and then multiplying by the slope of the derivative to get the delta. Delta is applied to all weights similarly and you can repeat the process propagating through the network.
Isn't Backpropagation just Setting right values in those parameters (Mathematically translated Biological functions)
Lol why make it so complex to understand?
@@yashaswikulshreshtha1588 Alas you cant just plot the inputs to the outputs and algebraically derive the weights. The only way to find the weights is to slide the weights into place... Or just use a python library. That's a lot easier.
Congrats on 8k!!
I don't know why you have so little subscribers... Your work is top quality!
Most probably because he rarely uploaded
13.5k subs this year-- I can't wait to see that grow exponentially in 2022 :-)
0:41 “im not big brain”
In the same year: “3d Graphing in a 2d Calculator (Desmos)
Yep, all checks out!
Excellent video! Amazing quality! Liked and subbed
Ffs this is so nuts and cool you sure know how to get the most out of Desmos
You are probably the most amazing and underrated RUclipsr I've met. I really hope you gain more subscribers, all your videos are a piece of art!
Agreed!
It looks fun and scary at a same time.
Now, train it. In Desmos.
bro you are awesome, new suscriber i know you are gonna get far
this is really cool! hurts my brain, but cool!
And like that, 42k inodes cried out in pain
These videos have been popping up and all I can think of when I see "NN in desmos" is "why?" So I click and Im laughing all video lol
im not big brain then proceeds to do some math that looks extremely complicated
Mean channel growth g
this is phenomenal content
I watch these videos to feel smart even though I have no idea what's going on
Your brain is so big it has an atmosphere
I am gonna start digital protest for not giving this channel enough fame and recognition. You're kind of me because I like to do mind boggling, mind shattering stuff and seeing you not on top because of youtube I am sad.
Now Desmos is going to your house.
ur so underrated man wtf such good content
Like your fancy words magic man
This is a fantastic video
this guy is on another level
Wowwww this is such an interesting video man
The quality of your videos are insane, the content and the presentation. What video editor do you use?
How do you find all this info about Desmos? I didn't even know you could extract the x and y values of a point until I saw it in someone's Desmos file. Their documentation only adds a little onto what's obvious.
This just got recommended to me, video could get a ton of views pretty soon 🙂
0:41 "I'm not big brain"
jesus dude. why. i love it 😻
two months later : skynet created in desmos
Finally, a non-python machine learning video.
I am losing my brain
Perfect music choice, lmao
Hey Joghn, did you know that you can doodle on a pixel grid in Desmos?
damm mans went from 100 to 10k in a year good stuff
"i'm not big brain"
well, WHAT THE FUCK THEN-
This is seriously underrated
Banger every time.
I’ve had success classifying numerics with an RBF network, which might be basic enough to be used like this.
It seems like you're using sigmoid activations for the hidden layers; ReLU would mostly likely work better for hidden layers.
Cool, but can you also do it in Minecraft?
Oh and the other way around, have Desmos run Doom?
wait when did this jump up from 130 to 5.6K?
epic music
isn't that the theme of a demonic cat?
it fits tho
Next up, Desmos BF interpreter
Neural what now?
After seeing this I am not calling myself programmer anymore
Underrated.
Great work!
Yeah this is why I'm not a data science major
Lol congrats on the boost of subscribers.. Ya got me in this vid 👍
Maybe this is not a big-brain video but it`s impressive and inventive
Subbed for big brain time
I got this video after googling how to make a line is desmos. I’m fucked
we are at a point in time where i cannot tell if he talks in a monotone way or if he has a pretty good voice synth
you're a nice little gem on youtube
I have no idea what any of this means or is used for, but it was an interesting 5 minutes
Same man
Damn dude...and I thought I was enthusiastic about Math and CS....😶
next up....
"Recreating a human consciousness in Desmos"
you just gained a new sub
good stuff man
Jesus u made this when you only had 100 subs
Immediately subscribes
alright now do captcha with desmos
Though didn’t understand anything but you are coolll
So essentially, you made an AI and then trained it in Desmos? (I haven’t watched the full vid yet)
sick video
Now make a 6502 emulator in Desmos
i tested CNNs a few years ago, and it reached 98% accuracy
Software for the video?
When's the desmos API coming out loool
So cool
1:40 yeah I didn’t get any of that
Very cool 😎
Ah, cool!
ur kidding? it's 14k now
"I'm not big brain" so that was a lie.