Here is a quick *summary* of this video: -SVM can be used to do *binary* classification -SVM finds a *hyper-plane* (line in 2d, plane in 3d, etc) that separates its training data in such a way that the distance between the hyper plane and the closest points from each class is maximized -once SVM finds this hyper-plane, you can classify new data points by seeing which side of this hyper-plane they land on -SVM can only be used on data that is *linearly separable* (i.e. a hyper-plane can be drawn between the two groups) -Fear not though, as a common way to make data linearly separable is to map it to a *higher dimension* (but beware, as this is computationally expensive). -You can map it however you want, but there are established ways to do it, they are called *Kernels* . By using a combination of these Kernels, and tweaking their parameters, you'll most likely achieve better results than making up your own way :P -The really cool thing about SVMs are that you can use them when you have *very little data* compared to the number of features each of your data points has. In other words, when the number of data to the number of features per data ratio is low. Normally when this ratio is low, you experience overfitting, but since SVMs only use a few of your data points to create the hyper-plane in the first place, it doesn't really care that you give it such little data. Note however that accuracy of predictions is reduced when you use very little data. -SVMs simply tell you what class a new data point falls in, *not the probability* that it's in that class. This is of course a disadvantage. Thanks for such a fun, engaging, simple, yet *informative* explanation of SVMs! Really enjoyed watching this!
whoever came up with this Support Vector Machine method is a fucking genius! To try to convert a seemingly unsolvable situation to a familiar solvable situation and then apply the traditional solution. Such a simple concept but benefited so many industries. Salute. Wish I could be like the person.
MOST USEFUL, INFORMATIVE video I've come across yet for answering the question "WHAT IS a Support Vector Machine?" :-) So many OTHER videos try to tell you merely HOW to USE SVMs, WITHOUT actually DEFINING them; this is ever-and-always a clear indication of LACK OF GENUINE "UNDERSTANDING", because all their focus is on only on the "HOW"... IN SHARP (and HAPPY) CONTRAST, YOUR video appears genuinely PLEASED to START with an explanation of WHAT "Support Vectors" ARE, and HOW the term "Support Vector Machine" even got DERIVED ! FANTASTIC ! NICELY DONE. MORE grease to your elbows ! -Mark Vogt | Fellow Data Scientist/Consultant/Solution Architect in Chicagoland area...
The most succinct and beautiful explanation of SVM I have found! I was struggling to grasp the basics. Thank you so much for creating such a wonderful tutorial! :)
Wow! I spend many hours trying to understand what I have learned in classes... so many words and logic functions but no big picture in my head that helped me to understand why and how I use it. but you simplyfied it so nice with simple storytelling, pratical selfexplaining pictures and videos that give me a good picture why and how I use it. Thank you so much. Great work! ;-)
Thank you Rene I'm glad you enjoyed the video. 😁. Yeah the reason I made this video was to make these very hard topics easier to grasp especially for people who are just starting out in the field of machine learning.
Thanks for using the cat/dog identification example to explain the concept of SVM. After watching many videos, I came across the right one that gave me a basic idea of SVM.
Thank you Aditya G :). I am really glad you enjoyed the video and that it made sense to you. If you have anything that you want me to make a video about, just use the hashtag #augmentedstartups with your comment and I'll see what I can do. If you haven't already, check out my channel page to see all of the topics I've covered so far www.augmentedstartups.info/RUclips or check out our courses here www.augmentedstartups.com/store or you can consider becoming a member of Augmented Startups and get access to Advanced Tutorials :) ruclips.net/channel/UCFJPdVHPZOYhSyxmX_C_Pewjoin I look forward to seeing you around! 👊
TensorFlow is a free and open-source software library for machine learning. It can be used across a range of tasks but has a particular focus on training and inference of deep neural networks. Tensorflow is a symbolic math library based on dataflow and differentiable programming.
This is an icy cool explanation of a very tough concept to grasp, especially for beginners like me. Thank you so much for making this. Saves so much time and frustration.
I want to ask the difference between "features" and "dimensions". Because there are contradictions in the advantages and disadvantages of these two things. Also, can you tell which example is a feature or dimension in the image classification. Hope that I got the answer, thank you
Really intrigued by the way you teach these topics like they're nothing. Could you help me with how to get started in machine learning in python? I know how python 3 works.
The 2nd point of Advantage(5:08) and 1st point of dis-advantage(5:39) are looks similar. Which means in advantage you specified that SVM perform well even though the number of dimension or features is greater then number of samples. But in dis-advantage it is stated that SVM will perform poor if #features > #samples. Isn't looks like contradictory ? is my understanding right?
@@Mustafa-jy8el SVM can be used when the no of datapoints < the no of features or no of dimentions or variables when this is not the case it is a disadvantage. I think in the video there is a confusion.
If anybody is searching for the Oviedo tip, this is the tesis: Modelización y análisis de la calidad del aire en la ciudad de Oviedo (norte de España), mediante los enfoques PSO-SVM, red neuronal MLP y árbol de regresión M5
Thank you for such an insightful video. I finally learned what SVM is. Can you please tell me which software do you use for video editing? Thanks for all your help
Thank you Shreyas Chaturvedi :). I am really glad you enjoyed the video. If you have anything that you want me to make a video about, just use the hashtag #augmentedstartups with your comment and I'll see what I can do. If you haven't already, check out my channel page to see all of the topics I've covered so far www.augmentedstartups.info/RUclips or check out our courses here www.augmentedstartups.com/store or you can consider becoming a member of Augmented Startups and get access to Advanced Tutorials :) bit.ly/Join_AugmentedStartups I look forward to seeing you around! 👊
In the advantages of SVM you mentioned it is useful when number of dimensions is greater than the samples, and in the disadvantages you said it has poor performance when number of features is greater than number of samples. What is the difference between the 2 (Dimension and features) ?
So if SVM has the disadvantage of not giving the probability, it can't be used for predictions right? What about using SVR for prediction? Could you do a video about SVR?
![FNAF Into the Pit has A LOT of Unused Graphics | LOST BITS [TetraBitGaming]](http://i.ytimg.com/vi/6YhZCRdipl4/mqdefault.jpg)
⭐ If you enjoy my work, Id really appreciate a Coffee😎☕ - augmentedstartups.info/BuyMeCoffee
Here is a quick *summary* of this video:
-SVM can be used to do *binary* classification
-SVM finds a *hyper-plane* (line in 2d, plane in 3d, etc) that separates its training data in such a way that the distance between the hyper plane and the closest points from each class is maximized
-once SVM finds this hyper-plane, you can classify new data points by seeing which side of this hyper-plane they land on
-SVM can only be used on data that is *linearly separable* (i.e. a hyper-plane can be drawn between the two groups)
-Fear not though, as a common way to make data linearly separable is to map it to a *higher dimension* (but beware, as this is computationally expensive).
-You can map it however you want, but there are established ways to do it, they are called *Kernels* . By using a combination of these Kernels, and tweaking their parameters, you'll most likely achieve better results than making up your own way :P
-The really cool thing about SVMs are that you can use them when you have *very little data* compared to the number of features each of your data points has. In other words, when the number of data to the number of features per data ratio is low. Normally when this ratio is low, you experience overfitting, but since SVMs only use a few of your data points to create the hyper-plane in the first place, it doesn't really care that you give it such little data. Note however that accuracy of predictions is reduced when you use very little data.
-SVMs simply tell you what class a new data point falls in, *not the probability* that it's in that class. This is of course a disadvantage.
Thanks for such a fun, engaging, simple, yet *informative* explanation of SVMs! Really enjoyed watching this!
Thanks..
That is well summarize, thank you :).
Thanks bro
This is cool than video thanx man
Thanks
whoever came up with this Support Vector Machine method is a fucking genius! To try to convert a seemingly unsolvable situation to a familiar solvable situation and then apply the traditional solution. Such a simple concept but benefited so many industries. Salute. Wish I could be like the person.
⭐ Haha yeah he or she is genius!! BTW if you enjoy my work, Id really appreciate a Coffee😎☕ - augmentedstartups.info/BuyMeCoffee
I learnt more in this video than two months of classes.
It means a lot. Thank you for the comment. I'm glad I could help 😊
mi2
Exactly 😂😂😂😂
You are in wrong class
⭐ Thanks, Also if you enjoy my work, Id really appreciate a Coffee😎☕ - augmentedstartups.info/BuyMeCoffee
MOST USEFUL, INFORMATIVE video I've come across yet for answering the question "WHAT IS a Support Vector Machine?" :-)
So many OTHER videos try to tell you merely HOW to USE SVMs, WITHOUT actually DEFINING them; this is ever-and-always a clear indication of LACK OF GENUINE "UNDERSTANDING", because all their focus is on only on the "HOW"...
IN SHARP (and HAPPY) CONTRAST, YOUR video appears genuinely PLEASED to START with an explanation of WHAT "Support Vectors" ARE, and HOW the term "Support Vector Machine" even got DERIVED ! FANTASTIC !
NICELY DONE.
MORE grease to your elbows !
-Mark Vogt | Fellow Data Scientist/Consultant/Solution Architect in Chicagoland area...
Took Linear Algebra and _just_ learned what all that "kernal" stuff was about. Thank you!
The most succinct and beautiful explanation of SVM I have found! I was struggling to grasp the basics. Thank you so much for creating such a wonderful tutorial! :)
Wow! I spend many hours trying to understand what I have learned in classes... so many words and logic functions but no big picture in my head that helped me to understand why and how I use it. but you simplyfied it so nice with simple storytelling, pratical selfexplaining pictures and videos that give me a good picture why and how I use it. Thank you so much. Great work! ;-)
Thank you Rene I'm glad you enjoyed the video. 😁. Yeah the reason I made this video was to make these very hard topics easier to grasp especially for people who are just starting out in the field of machine learning.
Ahh... You show Majin Boo under the Margin... Smart... And really wonderful Explanation of SVM. keep up the good work.
I'm glad there are some Dbz fans out there 😁. Thanks for the comment.
Thanks for the simplified explanation, it makes learning fun, you are my academic hero
Love these comments😁. I'm glad I could help and make learning fun! Thank you.
This is better than I was hoping for! Thanks so much for making videos that easily summarize the important parts of my uni papers!
I'm glad I can help :)
Thanks for using the cat/dog identification example to explain the concept of SVM. After watching many videos, I came across the right one that gave me a basic idea of SVM.
Thank you Aditya G :). I am really glad you enjoyed the video and that it made sense to you. If you have anything that you want me to make a video about, just use the hashtag #augmentedstartups with your comment and I'll see what I can do.
If you haven't already, check out my channel page to see all of the topics I've covered so far www.augmentedstartups.info/RUclips or check out our courses here www.augmentedstartups.com/store
or you can consider becoming a member of Augmented Startups and get access to Advanced Tutorials :)
ruclips.net/channel/UCFJPdVHPZOYhSyxmX_C_Pewjoin
I look forward to seeing you around! 👊
Thanks a lot sir।
Very helpful video for me .
Love from republic of india. ❤️❤️❤️
You are most welcome :)
I see what you did with that "Margin" Buu :D.
Anyway thanks this was easy, clear and isn't boring like most other guides
⭐ Haha, Thanks, Also if you enjoy my work, Id really appreciate a Coffee😎☕ - augmentedstartups.info/BuyMeCoffee
TensorFlow is a free and open-source software library for machine learning. It can be used across a range of tasks but has a particular focus on training and inference of deep neural networks. Tensorflow is a symbolic math library based on dataflow and differentiable programming.
Very Helpful to understand the Basic concept. Thank You.
Awesome you added Margin Buu, I have never thought about it like that.
Had to throw in the dbz reference 😁
Your explanation was phenomenal. No one could possibly explain it in simpler terms.
+Bennet Eapen thank you, glad you enjoyed it :)
One of the best videos I have seen on RUclips to date. Given a perfect intuition and explanation on SVMs!
Thank you so much 😁. Please share this video if it was helpful, I'd really appreciate it
@@Augmented_AI Definitely! :)
BTW, do you know any good APIs which provide Traffic flow History data?
@@padisalashanthan98 nothing as yet. But I'll look into it
@@Augmented_AI Thank you very much!
Great Introduction..your usage of visual aids is just fantastic!!
It's really helpful to understand with some real time example! Thanks!
The first point discussed in advantages contradicts the one discussed in disadvantages. Please clear this!!
This is very helpful. Thanks for creating this valuable content!
Thank you I'm so glad you enjoyed it 😊
This is an icy cool explanation of a very tough concept to grasp, especially for beginners like me. Thank you so much for making this. Saves so much time and frustration.
Thank you!well explained
nice.. you make it easy to understand the concept of svm
Glad you enjoyed it 😁. What would you like to see next
Thanks and lots of love from INDIA 😍
Thank you Sachin. Really appreciate it 😁
Thank you sir for teaching in an easy and understandable way
I love your approach on teaching things that should be made fun learning :)
Very intuitive. Explained SVM so clearly.
I want to ask the difference between "features" and "dimensions". Because there are contradictions in the advantages and disadvantages of these two things. Also, can you tell which example is a feature or dimension in the image classification. Hope that I got the answer, thank you
Awesome, I can finally understand what the SVM is!
RIP Grumpy Cat :) You were the GOAT
i live for this kind of explanation. Cute and easy to understand. Thanks!
Yeah. 😁 Learning should fun right?
thats the best explanation i heard yet
Amazingly easy explained!
you are a genius really genius much thanks, may Allah bless you
🙏
This video helped in clear conceptual understanding on non linear SVM. Thanks for uploading
Really intrigued by the way you teach these topics like they're nothing. Could you help me with how to get started in machine learning in python? I know how python 3 works.
The 2nd point of Advantage(5:08) and 1st point of dis-advantage(5:39) are looks similar. Which means in advantage you specified that SVM perform well even though the number of dimension or features is greater then number of samples. But in dis-advantage it is stated that SVM will perform poor if #features > #samples. Isn't looks like contradictory ? is my understanding right?
yeah, can someone explain that? Isn't it that the #features == #dimensions?
I'm having the same confusion. Can someone help?
@@Mustafa-jy8el SVM can be used when the no of datapoints < the no of features or no of dimentions or variables
when this is not the case it is a disadvantage. I think in the video there is a confusion.
I was confused by this too
same
You deserve a lot more subscribers. Awesome explanation :)
Thank you so much Amresh, I really appreciate the comment. :) The subscribers will come soon 😎
Great video! Good job!
Thank you so much Thomas 😁
neat and precise, thanks for your explaination
You are most welcome. What would you like to see next?
@@Augmented_AI how big Reinforcment Learning Projects like for Dota 2 are aproached :)
Thank you, that helped me a lot calming down before the exam tomorrow on Machine Learning!
Great explanation. The use of visual examples make it easy to understand SVM
The Best Explanation, I ever hear about the support vector machine..
Simple and easy explanation of SVM. Thank you.
Your videos are always amazing and so well-explained.
Stop scrolling to comments. pay heed to what he is teaching. he is awesome
This video was so good! Thanks to you I'll pass my data science class!
That's really great to hear Jack! 😀 I'm really glad that these videos could help you.
If anybody is searching for the Oviedo tip, this is the tesis: Modelización y análisis de la calidad del aire en la ciudad de Oviedo (norte de España), mediante los enfoques PSO-SVM, red neuronal MLP y árbol de regresión M5
Best video
Love from india
great video . Thank you
Thank you Hk 😁
i can easily understand how svm works! thx for the video!
Your teching is very good. Thank you sir for teaching us in easy way.
You are most welcome Ishikia :)
Wonderful explanation.. thank you so much..
+DINESHKUMAR MURUGAN thank you so much for the support :)
Good examples
This was very clear and helpful, thank you!
Excellent video !
Great channel for educational videos, the best, very interesting !!
Good video! crisp! it would be great if you make v2 of this video touching "soft margin" SVM as well
Excellent video. Thanks for showing the applications of such powerful tool.
+filick82 thank you, I really appreciate it :)
Cool and very useful.
Thank you Manuel 😁
Awesome video bro!
Thanks man 😆😁🤟
Very well explained! 👍
Thank you Rishabh 😁
6:22
Holy damn! I know a guy from nearby there xd
A very excellent explanation..... And efficient use of visualisation
What is Twin support vector machine (TWSVM) and how it is differ from SVM pls tell me
Thank you very much.
YOU INSANE!! Augmented Startups!
Aweeeeeeeeeesomeeee...one of the best videos on SVM's
sir please make exploratory video on gender text idenfication features and techniques
It was pretty pretty useful! Thank you so much!
+Valeria Pérez - Cong thank you for the comment :) I really appreciate it.
Awesome Explanation covered and explained brilliantly.
Thank you Aman 😁. Really appreciate the feedback
thank you
You're welcome🙂
Wonderful explanation sir!
Thank you Kaisher :)
Awesome video! You have a great style of teaching.
Thank you 😃😁
Thank you for such an insightful video. I finally learned what SVM is. Can you please tell me which software do you use for video editing?
Thanks for all your help
I'm glad I could help you learn SVM 😁👍. Sure I use video scribe for it.
Thanks for letting me know 😊
Thanks a lot sir.... :)
very well explained! thank you :)
Thank you Shreyas Chaturvedi :). I am really glad you enjoyed the video. If you have anything that you want me to make a video about, just use the hashtag #augmentedstartups with your comment and I'll see what I can do.
If you haven't already, check out my channel page to see all of the topics I've covered so far www.augmentedstartups.info/RUclips or check out our courses here www.augmentedstartups.com/store
or you can consider becoming a member of Augmented Startups and get access to Advanced Tutorials :)
bit.ly/Join_AugmentedStartups
I look forward to seeing you around! 👊
thanks dude
You are most welcome :)
Nice video! All I have to add is, "Margin Buu" :p
Incredible explanation sir.
Really good one.. keep up the good work
Thank you Aakash 😄
this video is very helpful to understand support vector machines
+Shaik Shabeena Begum thank you I'm glad it was easy to understand :)
awesome
i love it
Thank you Raunak. What would you like to see next from us?
neatly explained !!
Thank you Vindhya 😁👍
Thanks a lot !
In the advantages of SVM you mentioned it is useful when number of dimensions is greater than the samples, and in the disadvantages you said it has poor performance when number of features is greater than number of samples. What is the difference between the 2 (Dimension and features) ?
same, I'm still confused about it too. have you gotten the answer?
Thank you sir
Very good! Just nickpicking: "ear geometry" is not a parameter. Ear area would be more parameter-able.
Mejor canal educativo, buen contenido excelente
Love this vid!!!! Thank youuu ❤️❤️❤️
I'm glad you do 😊. Thank you Vaidehi
A great explanation. Could you suggest where we could learn these type of animations for teaching ? Thank you.
thanks for easy explanation.
I'm having a little trouble in understanding what kernel trick is. Could you please explain a bit more about it?
So if SVM has the disadvantage of not giving the probability, it can't be used for predictions right? What about using SVR for prediction? Could you do a video about SVR?
Amazing!! And that cat-dog image is killing me 😂
I'm glad you enjoyed it 🤣. What would you like to see next?
⭐ Haha, Thanks, Also if you enjoy my work, Id really appreciate a Coffee😎☕ - augmentedstartups.info/BuyMeCoffee
The best explained video onSVM
Thank you Bilal, I really appreciate it :)
Amazing video! Thank you for uploading.
I'm glad you enjoyed it :)