Dude that was a fantastic explanation! and the video illustrations were excellent! and you really went over and above with the reference links for deeper studies! subscribed! keep up the good work! :D
Sorry I'm super late to this. TH-cam didn't notify me of this amazing comment. Thanks a ton! We can chat better on Discord since I'm more active there. Link in the latest video description
I like the format of "logistic regression - the math you should know" better, I think the intro here is a little bit long and I think the viewers of this video will know a bit about ML but are more interested in the details of boosting (speaking for myself at least) Thank you ! keep it up !
Sir, just one question.... where you learn maths behind the machine learning algorithms... I am trying really hard to find courses about mathematics but i failed.... Where i can find resources to learn mathematics behind machine learning algorithms...
there is a channel called statquest you can have some decent idea math in that. MIT has a fab course called Artificial Intelligence by Patrick wilson they introduce you to some math there. And there are lot of medium articles where you can see the math. You will have to dig some more deeper. Machine learning algos are not built on one single ideas. Like in decision tree and even in ada boost you have an idea called gini score and all . Its a measure of entropy . And entropy is a information theory based ideas. Librarys are the most easiest way to approach this if you start understanding the math then there are lot of dependency. Also a decent idea of statistics , propablity , calculus can help you understand the ideas better. Because this algos are built on top of it.
Yup. :) I'm trying a different approach with more visuals and easier explanations (without losing detail). So it took longer. I'd actually been working on this almost every day for the last month after work. Next step is to probably decrease the video length to make it more palatable (?) - not too sure. But will see how it goes. Thanks for the support! :)
That's a good question. If I know the topic I'm looking for, I'd just Google it (like Xgboost). For "history" of boosting though, I'd also try to find college lecture material. They have a good explanation at a high level, but I'd dig into their references for more info. Apart from that there is arxiv sanity and social media that I use for more trending research (explained this more in my video on "how to keep up with AI research. Check it out)
Awesome, I will use XGboost for some classification problems. What's program do you use to make your videos? I would like to learn about it, but I don't have a clear path to learn those skills on Internet.
XGboost can also be used for regression too. Since the base weak learner is a decision tree. I use Camtasia studio for creating these videos. It's great for recording your screen. And if you play around long enough with it, you can create decent animations.
You lost me when you didn't explain what a gradient is or how it differs from a weight. That made the rest unintelligible. I hope you can improve this.
Yea. This video is from 4 years ago. I have definitely improved over time. But to answer your question in a nutshell. Weight = parameter, gradient = change in said parameter
You got my respect man. I think this is the only video that actually cared enough to define what strong and weak learners are.
Thanks! Tried to get deep with this one
Dude that was a fantastic explanation! and the video illustrations were excellent! and you really went over and above with the reference links for deeper studies! subscribed! keep up the good work! :D
Exactly!
Sorry I'm super late to this. TH-cam didn't notify me of this amazing comment. Thanks a ton! We can chat better on Discord since I'm more active there. Link in the latest video description
Thanks Mr. Code Emporium you are as good as 3 blue one brown at explaining the difficult.
Great explanation. Had no idea what boosting was and this video just demystified the whole thing. Big up!
What a great explanation and fantastic work! Appreciated those references!
Amazing quality of production! Appreciate your effort!
one of the BEST videos for this subject
Thank you so much!
Great video! I'm using this to research for a video I'm working on now!
I am honored! Can’t wait to see it!
This is beautifully explained!
Great simplified explanation 👍 Thanks 😊
I like the format of "logistic regression - the math you should know" better, I think the intro here is a little bit long and I think the viewers of this video will know a bit about ML but are more interested in the details of boosting (speaking for myself at least)
Thank you ! keep it up !
Yeah. I'm working on getting to the point much quicker. Thanks for the feedback!
Great Explanation!
Thanks so much! :)
Thanks man. Great explanation as always. Wish you all the best!
Thanks! This is helpful.
Awesome explanation 👏 👌
Very nice
So freaking amazing!
What an amazing video.
This video is GREAT!
you are GREAT!
crazy good quality video, thank you!
thank you man, this was amazing.
Thank you 🙏
Amazing video!
Thanks!
great explanation thanks bro
You did not explain why increasing the sample weight makes the next iteration focus on the misclassified samples.
Excellent! Thanks!
why it select weak model suppose if I get 95% accuracy in first model and its is selection weak model that is having 65% accuracy why?
So great, thanks man!
Sir, just one question....
where you learn maths behind the machine learning algorithms... I am trying really hard to find courses about mathematics but i failed....
Where i can find resources to learn mathematics behind machine learning algorithms...
there is a channel called statquest you can have some decent idea math in that. MIT has a fab course called Artificial Intelligence by Patrick wilson they introduce you to some math there. And there are lot of medium articles where you can see the math. You will have to dig some more deeper. Machine learning algos are not built on one single ideas. Like in decision tree and even in ada boost you have an idea called gini score and all . Its a measure of entropy . And entropy is a information theory based ideas. Librarys are the most easiest way to approach this if you start understanding the math then there are lot of dependency. Also a decent idea of statistics , propablity , calculus can help you understand the ideas better. Because this algos are built on top of it.
I think Andrew Ng's ML videos might come handy too
SUBSCRIBED!!!!!
NO REGRETS! THENKS!
But how does a model "focus more on a problem to make sure it gets it right"? What does that mean?
So finally u uploaded a video 😂
I like ur explanation very much
Yup. :) I'm trying a different approach with more visuals and easier explanations (without losing detail). So it took longer. I'd actually been working on this almost every day for the last month after work. Next step is to probably decrease the video length to make it more palatable (?) - not too sure. But will see how it goes. Thanks for the support! :)
This has 558 likes and cat videos have millions of likes. The world is not a fair place!
A cruel world we live in :)
Hi, stupid question, but how you find the research papers exactly because they're such great! Thx for the gorgeous explanation, helped me a lot!
That's a good question. If I know the topic I'm looking for, I'd just Google it (like Xgboost). For "history" of boosting though, I'd also try to find college lecture material. They have a good explanation at a high level, but I'd dig into their references for more info. Apart from that there is arxiv sanity and social media that I use for more trending research (explained this more in my video on "how to keep up with AI research. Check it out)
Which program do you use to create the videos?
nice
Who are you? Where were you all my life? You are amazing! Do you have Pateon?
Awesome, I will use XGboost for some classification problems.
What's program do you use to make your videos? I would like to learn about it, but I don't have a clear path to learn those skills on Internet.
XGboost can also be used for regression too. Since the base weak learner is a decision tree.
I use Camtasia studio for creating these videos. It's great for recording your screen. And if you play around long enough with it, you can create decent animations.
You lost me when you didn't explain what a gradient is or how it differs from a weight. That made the rest unintelligible. I hope you can improve this.
Yea. This video is from 4 years ago. I have definitely improved over time. But to answer your question in a nutshell. Weight = parameter, gradient = change in said parameter
@@CodeEmporium ok thanks!