One of your greatest videos sir. Hands down... Perfect timing perfect explanation... Wow simply amazed.. I downloaded the video so that I have a copy of it. Just in case UA-cam is non existent in a zillion years
The best ever video on Logistic Regression ever seen. SIr, keep sharing tutorials like this, very helpful in making deep understanding and implementations.
Rather then picking the points randomly, what if we linearly iterate over the array of points? I think by doing this we can improve our model accuracy because in case of picking a random point it may happen that random function keep on picking some few points every time. Sir what do you think upon this? By the way, your explanation was really good
I hope you could create the "Maximum Entropy Markov Model" which training based with logistic regression but the test with Viterbi maximum entropy is logistic regression with more than one class
Halfway into the video, I thought: "This is the Khan Academy of Machine Learning!" Really great explanations. One thing you didn't drive home as much, is that I could just code this up and it would work. I imagine this is still true though (for perceptron and logistic).
Thanks! Yes, I plan to start making GitHub lavs with videos, just need to get around it. :) If you have any code you'd like to share, lemme know and I'll link it!
Sir, you are awesome, I have just started learning Data Science and your explanations really clear the fog from the picture. Thank you so much for putting such wonderful content on internet. If you have any paid course on Data Science for beginners like me then please do share the link of that in reply of this comment. And once again, your explanations are just great, loved it❤
In some videos about perceptron, lecturers say "if our total error is 0 (linearly separable) then perceptron algorithm finds that linear classifier after some iteration." Perceptron decreases the error but it may not be 0. And if perceptron finds minimum error linear classifier, we can call that as linear classifier as well. Not just a perfect one. Can you correct me about that Luis? Also thanks for your work, its a sound introduction.
Please make the separate video on logistic regression that how it is different from linear. You mixed it with perceptron algorithm. You explained linear algorithm very well but I unable get clear understanding about logistic through this video.
The gradient descent challenge might need some further help on "What is Gradient Descent?" I know there are lots of videos on this online, but do you have you you like or you produced?
Yeah, I need to do some serious material on gradient descent. I have something on a video called "A friendly introduction to deep learning", if you'd like to take a look.
Yes great questions, I think it's because the log error, or the logit, which is the function sending everything to [0,1]. Although the strangest thing for me is that it's called logistic regression, even though it is not regression, but classification. Linear regression can use a quadratic error, or an absolute value (almost linear) error, so the reason it's called linear is because the output is a line.
11:11 I thought the epoch is how many times we scroll through *all* the points, not how many points we pick to repeat the process of slightly moving the line. Am I wrong ? Thanks for help BTW, why is it called *Perceptron algorithm* ? It has nothing to do with perceptron ?
i want implementation of the following pseudo code : Step I: Start with a random line of equation ax + by + c = O Step 2: Pick a large number. 1000 (number of repetitions, or epochs) Step 3: Pick a small 0.01 (learning rate) Step 4: (repeat 1000 times) Pick random point from If point is correctly classified Do nothing If point is incorrectly classified Add ± 0.01 to a Add ± 0.01 to b Add ± 0.01 to c help me
Definitely! HEre it is: github.com/luisguiserrano/manning/tree/master/Chapter_5_Perceptron_Algorithm In that same repo github.com/luisguiserrano/manning/ I have many others from the videos
Hi Luis, great video! The way you explained these concepts was nice and easy to understand. I decided to implement the basic perceptron algorithm in python, feel free to check it out here: github.com/nenslen/perceptron I tried to use the same terminology as you did in the video (eg. red and blue points), so it should be easier to relate it to your explanation. If you end up taking a look, any comments or feedback are appreciated, thanks!
One of your greatest videos sir. Hands down... Perfect timing perfect explanation... Wow simply amazed.. I downloaded the video so that I have a copy of it. Just in case UA-cam is non existent in a zillion years
The best ever video on Logistic Regression ever seen.
SIr, keep sharing tutorials like this, very helpful in making deep understanding and implementations.
Your tutorial is very easy to grasp. Best channel on machine learning with good graphic demonstration. Thank you for your hard work
You sir are a great teacher! The way you explained that was simply amazing! That was so engaging!
Just saw this. Wow. I'm looking forward to seeing ALL of your vids. Thank you for your kindness and time to share with the world.
Just want to leave a comment so that more people could learn from your amazing videos! Many thanks for the wonderful and fun creation!!!
Thank you! :)
This is the best channel for Machine learning. Best to understand.
you are the best in the business of teaching. I came across your content couple of days ago and now I am addicted.
Cheers to you men. I just opened a beer to celebrate. I finally understand what Gradient Descent is how it works and why . Thanks Luis
Thank you very much for depicting the mathematical concepts into very simple graphs! Unforgettable!
You deserve more subscribers. All my professors should be at your level. Great video, I learned too much!
Rather then picking the points randomly, what if we linearly iterate over the array of points? I think by doing this we can improve our model accuracy because in case of picking a random point it may happen that random function keep on picking some few points every time. Sir what do you think upon this?
By the way, your explanation was really good
Amazing Video! This video explains logistic regression in a layman language which helped me understand the algorithm in depth! Thank-you so much!
Best tutorial on perceptron out there! Thank you
Really really great. I looked for this explanation for months
Amazing, this really was as friendly as this topic can get.
The best #MachineLearning videos are by Luis Serrano. Respect + Thank you, Maestro!
I hope you could create the "Maximum Entropy Markov Model" which training based with logistic regression but the test with Viterbi
maximum entropy is logistic regression with more than one class
Halfway into the video, I thought: "This is the Khan Academy of Machine Learning!" Really great explanations. One thing you didn't drive home as much, is that I could just code this up and it would work. I imagine this is still true though (for perceptron and logistic).
Thanks! Yes, I plan to start making GitHub lavs with videos, just need to get around it. :)
If you have any code you'd like to share, lemme know and I'll link it!
love u for making my life so much easier. And adding fuel for passion in learning. This video helped a ton.
One of the best videos so far on the logistic regression..
Sir, you are awesome, I have just started learning Data Science and your explanations really clear the fog from the picture. Thank you so much for putting such wonderful content on internet. If you have any paid course on Data Science for beginners like me then please do share the link of that in reply of this comment. And once again, your explanations are just great, loved it❤
The best video I've come across to understand perceptron algorithm concept... And believe me I've tried many! :-)
In some videos about perceptron, lecturers say "if our total error is 0 (linearly separable) then perceptron algorithm finds that linear classifier after some iteration." Perceptron decreases the error but it may not be 0. And if perceptron finds minimum error linear classifier, we can call that as linear classifier as well. Not just a perfect one. Can you correct me about that Luis? Also thanks for your work, its a sound introduction.
Thank you very much for such a great in-depth explanation. Please continue uploading more and more videos.
Great video.Learnt the concept in an hour
Can you please please make a video on maximum likelihood estimation or expectation maximisation
Awesome tutorial. The best explaination on Logistic regression and perceptron alogorithm! Take a bow Luis!
Please make the separate video on logistic regression that how it is different from linear. You mixed it with perceptron algorithm. You explained linear algorithm very well but I unable get clear understanding about logistic through this video.
Very good Luis. I am new in your channel. Im here because your book in grooking is nice
thank you i am watching again and again I bought your book I am abeginner thank you again you and stammer made things very easy
The gradient descent challenge might need some further help on "What is Gradient Descent?" I know there are lots of videos on this online, but do you have you you like or you produced?
Yeah, I need to do some serious material on gradient descent. I have something on a video called "A friendly introduction to deep learning", if you'd like to take a look.
Thank for your excellent video. It was a real help!
I kind of wonder why it's called Logistic Regression. Because it uses a Log-Based Error? Did Linear Regression use a linear error?
Yes great questions, I think it's because the log error, or the logit, which is the function sending everything to [0,1]. Although the strangest thing for me is that it's called logistic regression, even though it is not regression, but classification.
Linear regression can use a quadratic error, or an absolute value (almost linear) error, so the reason it's called linear is because the output is a line.
Thank you so much. Its very easy to understand and your presentation skill are awesome. Sir would u please share the slides ??
Watching for 3rd time.. not because I couldn't understand, but for the beauty and simplicity of explanation
sir, your awesome the way u explain things are very simple and clear
Thanks Indratej! There's a whole deep learning course I taught here (free)! www.udacity.com/course/deep-learning-pytorch--ud188
@@SerranoAcademy sure i will check out sir thanks for sharing
when are you planning to finish Chapter 7 on Linear Regression of your book : Grokking Machine Learning
Sir, Your explanation is very well. It helped me a lot. Thank you so much. Please make some others videos of ML.
Excellent clear and informative explanation.
Amazing work, thank you!
Best video I've seen on Perceptron Algorithm hands down. Thank you Luis!
"Hello grandson, I have made cookies, love grandMa" ... that's obviously spam
Wow, this is absoulute gold.
Thank you for sharing it
Thanks for that video !
Great video thank you so much.
excellent video thanks for sharing!
You clever and clear my 5th grade son understands it.
Lovely, thanks a lot dear.
11:11 I thought the epoch is how many times we scroll through *all* the points, not how many points we pick to repeat the process of slightly moving the line. Am I wrong ? Thanks for help
BTW, why is it called *Perceptron algorithm* ? It has nothing to do with perceptron ?
you are right, I guess he is referring to batch not epoch
Wow I have been month try to understand ML thank you a lot
Very clear. Thanks.
Can you do a video on variational bayed and KL divergence?
Thanks for the lesson Luis :)
i want implementation of the following pseudo code :
Step I: Start with a random line
of equation ax + by + c = O
Step 2: Pick a large number. 1000
(number of repetitions, or epochs)
Step 3: Pick a small
0.01
(learning rate)
Step 4: (repeat 1000 times)
Pick random point from
If point is correctly classified
Do nothing
If point is incorrectly classified
Add ± 0.01 to a
Add ± 0.01 to b
Add ± 0.01 to c help me
Definitely! HEre it is: github.com/luisguiserrano/manning/tree/master/Chapter_5_Perceptron_Algorithm
In that same repo github.com/luisguiserrano/manning/ I have many others from the videos
🥰🥰😍😍@@SerranoAcademy
Great for developing intuition!
wonderful Thanks
Nailed it... Thanks
Sir, please explain ANCOVA & GLM
Amazing as always
Gracias por compartir este video!
The error is minimized as the derivative of the error function approaches zero.
Thank you, once again!
Thank you for this video. :)
Great content! Thanks so much :)
THANK YOU SO MUCH!!!
very nice video!
Hi Luis, great video! The way you explained these concepts was nice and easy to understand. I decided to implement the basic perceptron algorithm in python, feel free to check it out here:
github.com/nenslen/perceptron
I tried to use the same terminology as you did in the video (eg. red and blue points), so it should be easier to relate it to your explanation. If you end up taking a look, any comments or feedback are appreciated, thanks!
Nice...
I love you!!!!😍
Clever great!
❤