Gradient Descent From Scratch in Python - Visual Explanation
Vložit
- čas přidán 17. 04. 2023
- In this video we implement gradient descent from scratch in Python. Also I try to give you an intuitive and mathematical understanding of what is happening.
◾◾◾◾◾◾◾◾◾◾◾◾◾◾◾◾◾
📚 Programming Books & Merch 📚
🐍 The Python Bible Book: www.neuralnine.com/books/
💻 The Algorithm Bible Book: www.neuralnine.com/books/
👕 Programming Merch: www.neuralnine.com/shop
🌐 Social Media & Contact 🌐
📱 Website: www.neuralnine.com/
📷 Instagram: / neuralnine
🐦 Twitter: / neuralnine
🤵 LinkedIn: / neuralnine
📁 GitHub: github.com/NeuralNine
🎙 Discord: / discord - Věda a technologie
This is a very nice video explaining the mysteries behind gradient descent!
The greek "n" looking symbol is eta, as far as I am aware.
The greek letter "η" is called eta.
Very educational bro ,I love it❤
Aaaawesome, as expected 🔥🔥
Could you do logistic regression
Thank you so much, it really helped me out understanding the gradient descent :)!!!
Thank you very much for the video 🙌🏽
Great explanation and good visuals!
Super Awesome .. Wonderful explanation and i can understand this video has some prerequisite .. but what you can do is leave some references to people who are not graduated to this level
Thank you brother, this was immensely helpful
I’d recommend implementing a higher-order finite difference scheme for computing the gradient instead of hard-coding the analytic derivative for those who want to implement more general schemes.
Great Explanation..👍
Thank you very much
at the end, when you plotted 3 points, you only adjusted the x and y coordinates, but did not change the z coordinate. it was still z_function(0.7, 0.4)
Great video, thanks
Best tutorial ever seen
I spent around 490 hours learning calc 3 , now I gotta implement this algorithm, thx u for explanation
You counted the hours😮
@@unknown__user__800 yeah I actually did just to see how productive (I guess I can say ) I am
Thank you, very well explained 🙏
can you give me the code ?
very clean
Love this video
awesome bro
great tutorial
Amazing
thanks for the video. great job. I think you forgot to update the z function with the correct x and y values for positions 1-3.
Is that work with multiple features datasets?
16:16 Rollercoaster !!
10/10
thx
Hey can you please guide me, I am at begginer level , I am trying to built a python script or model which can detect wheather a pdf is an image one or non image , I got stuck at one problem: even if pdf is non image because it contains some type of stamps in it , it is considering it as image file and I am unable to segregate files , can you please suggest what should i do in this case.
Can anyone please teach me how to take the derivatives of this function: [sin(5x) * cos(5y)] / 5
Hello Sir,
When I execute: -
ax=plt.subplot(projection="3d", computed_zorder=False)
AttributeError: 'Axes3DSubplot' object has no property 'computed_zorder'
Can you please suggest how to resolve the problem?
Thank you
Awsm. When I run the code in colab, I get multiple plots. One plot for each iteration. It seems that plt.clf() is not clearing the plot. Can you please answer why its happening. Here is the code
def y_function(x):
return x**2
def y_derivative(x):
return 2*x
x = np.arange(-100,100,0.1)
y = y_function(x)
current_pos = (80, y_function(80))
learning_rate = 0.01
for _ in range (100):
new_x = current_pos[0] - learning_rate*y_derivative(current_pos[0])
new_y = y_function(new_x)
current_pos = (new_x,new_y )
plt.plot(x, y)
plt.scatter(current_pos[0], current_pos[1], color='red')
plt.pause(0.1)
plt.clf()
I got the same issue
did you find any solution?
I getting error in computed_zorder anyone is there any update
Calcuwho? The progging I get (mostly). The maths (While I do understand the underlying principles) somehow always produce soup when I try it (a form of dyslexia). The cool part is that these days we have new tools to extend ourselves mentally (and not at all in any naughty ways like some of you might be thinking despite 40% of the internet being dedicated to porn) such as Wolphram Alpha and ChatGPT. Using such tools allows us to extend the boundaries imposed by our own personal mental limitations.
To those of you who might be a bit more like me and fear treading the murky reef filled waters of higher maths remember that even if the mechanics elude you, a basic understanding is enough to use these tools to supplement ones capabilities. Even those who are well versed in math would be wise to make use of them given our roughly 15% error rates. Let the machine do the heavy lifting for the difficult and/or complicated bits.