Day 2-Forward Propogation, Loss Functions, Chain Rule Of Derivatives|Deep Learning Live
ฝัง
- เผยแพร่เมื่อ 18 พ.ย. 2024
- Enroll for free in the below link to get all the videos and materials
ineuron.ai/cou...
Live Deep Learning Playlist: • 5 Days Live Deep Learn...
Our Popular courses:-
Fullstack data science job guaranteed program:- bit.ly/3JronjT
Tech Neuron OTT platform for Education:- bit.ly/3KsS3ye
Affiliate Portal (Refer & Earn):-
affiliate.ineu...
Internship Portal:-
internship.ine...
Website:- www.ineuron.ai
iNeuron TH-cam Channel:-
/ @ineuronintelligence
Telegram link: t.me/joinchat/...
Please do subscribe my other channel too
/ @krishnaikhindi
Connect with me here:
Twitter: / krishnaik06
Facebook: / krishnaik06
instagram: / krishnaik06
This is the best video for me Krish has ever uploaded. Reason for being the best is because the volume and depth of knowledge he has covered here in 2hr can take a month or even an entire semester in a college DL course.
Best teacher in universe. More power and sccuess to you Krish.
You are one of the best teachers in this world. Thanks for the efforts!
By far the best tutorial on the DL topics I've ever seen. Joining now
I am totally speechless!! what an amazing explanation!! no one can teach data science concepts as you teach krish sir!!
Absolutely brilliant!! Inception level crazy made easy. I've been following your channel and live sessions Krish, haven't found a better teacher. Thank you.
What an amazing class, I thought deep learning is a Rocket science before. Amazing once again sir 💖
yes you are right
His teaching methods and concepts are absolutely brilliant!!! Thanks
the dashboard is not available. I am getting 404 page not found error
there are no notes available in community resources. pls provide the notes also
If you found . Please ping me .
It was a most important lecture. Thanks Krish sir
Great sir the way you teach all topics
Hi Krish 30:28,O2 directly not depended on O1 as you mentioned you are using activation fucntion so O2 came from Z2 and Z2 is the multiplication of O1 and 2nd layer weight
Super, thank you Krish.
Explanation is simply awesome through out. Extremely helpful! Thank you!
You truly are a wonderful teachers. Thank you sir
wow its just amazing krish Thank you so much for educating us at zero cost.
concepts are cristal clear and its boosts my confidence on Data science.
Really wonderful teaching.
excellent tutoring mr. @krish naik. please keep up this great community service. Thanks!
great teaching krish thank you
1:18:57 Why he not divide by n as we are calculating the Mean Squared Error.
Formula given : 1/(2) * (y - y_hat)^2
Formula should be : 1/(2*n) * (y - y_hat)^2
Yes it should be divided by n.... Missed that part
Wow yaar . Its extraordinary class . 😍😍😍😍Hats off to you Sir🙏
Thank u so much, sir! You r turning a complex topic into something so easily comprehensible and fun! I can easily understand the reason for different activation functions exist and why and when to use them. You are brilliant! Love ur videos so much!
Sir very nice explaination in a easy way. Chain rule explaination was amazing
With the bottom of my heart,Thank for you this
Excellent way of making us interested in DL techniques. Simple to the core methodologies . Good Krish . Keep it Going.
you have taken out the Deep learning fear out of me . thank Krish
Thankyou Krish this amazing
Awesome session krish....thankyou for great explanation :)
Thanks for a session @Krish
Excellently session by Krish sir
u are the best at all please make videos on web scrapping and dbms SQL no sql
This was a great session! Thanks for the content
Amazing Lesson sir!
Better than andrew ng. many of you will agree.
Every one has their own of teaching style .... bro
Krish during defining epochs to my model how can I know the number of epochs I need to reach global minima?
Sir , I did B.SC from statistics , and I'm interested to build a career in data scientist field . But sir I don't know any thing about technical skills and programming languages. I have seen iNeuron course which is going to start on 7th of may but sir due to some financial crisis I can't afford this time . Sir will you please guide me from where should I start learning so that I can get a job . It's just that I'll get some experience and financial stability too . By this my dependency on family for money wouldn't be a problem anymore for them and for myself also . Please sir guide me. Through which medium I can connect with you. Please sir help me out from this problem.
I don't know you got to connect with krish by now or not.
But i would like to say that you are already at the right place so don't worry, you have entered into the Gold mine(Krish Naik's YT channel).
First of all start by learning python,you can start with Krish's python playlist or code with harry python for beginners, then continue with different playlists of krish like statistics, linear algebra and some useful libraries like matplotlib, pandas and numpy then watch krish playlist for machine learning.
For practice projects you can also refer Siddhardhan's YT channel. He explains the projects in the best possible way for beginners.
After gaining some practice you can start with Deep learning playlist.
@@kakashihatake2052 Thank you soo muchh ...
Thank you so much Sir !
finished watching
Perfectly explained
Thankyou soo much sir! What an amazing session 👏👏
Thank you Krish.
10 out of 10 ratings for you always
Great material!!! Very well explained
brilliant teacher
wonderful class, Thanks
Hello...At 41:59 you are showing the weight update formula as:
partial L / partial w_1_new = w_1_old - learning rate * partial L / partial w_1_new
Should this be the following instead?
partial L / partial w_1_new = w_1_old - learning rate * partial L / partial w_1_old
Partial L/ partial w_1(old)
Too easiest by your way.. Thank you😇
Amazing session. You are really great sir and you have lot of patience to explain thank you so much sir.
As a student of jspm rscoe
Its good to see TEDx JSPM RSCOE Wrap on his Desk
For cost function is it 1/2or 1/n??
Sir notes was not available . link not working.Can you please check it sir.
At 01:17:08, when calculating loss for each record, how is mean abs error calculated? Shouldn't it be for all records and then backward propagation?
Awesome content
🥰 Great Lecture Krish
Thanks for the efforts
Thank you @krish
amazingly explained
It was amazing sir
i will rate you 10/10. And i took me 10 hours and to understand this video. Good Job.
10 hrs???? u should quit dl
@@whothefisyash yes. I am a slow learner but I learn new things everyday. That makes me stronger just like gym
@@whothefisyash Now I am a manager of Data Science team in a product based company. Hence slowness doesn’t matter proved
@@amitjena1556 good for u buddy
@@whothefisyash Thanks Man. All the best in career.
@krish Naik can we have the notebook that you shared in video?
Thankyou sir❤️❤️❤️❤️❤️❤️❤️🔥🔥🔥🔥
thank you
Thank you sir🙏🙏
Amazing session sir
hats of sir
brilliant session
25:17 and 42:51 - w(new) formula is mismatching. It will be learning rate multiplied by loss derivative of w(old) or w(new) ?
Yes, I also have same doubt. Please help me if you know the solution
W(old)
thank u sir
Amazing session sir.
I have done my bachelor's in applied mathematics but I didn't know an application of chain rule that is a basic topic in mathematics if that is about chain rule imagine what would be applications real analysis, complex analysis, topology ,functional analysis, linear algebra🤔
I actually have a doubt at 1:32:01, he says to calculate y^ we use sigmoid fn but as per my understanding, y^ is the predicted value and sigmoid fn is an activation fn, clearly the are not related by equality, rectify me if I am wrong !!
Y is the predicated value and y^ is the output
@@sheshugoudgaajarla2212 true that's wrong on my part, y^ is output value, still it's not equal to sigmoid fn right ?
@@abhirajsingh8138 Actually, we provide the y^ to the sigmoid function as the input. Then, the sigmoid function will get activated and convert/restrict the output in the range 0 to 1. It can be any value between 0 and 1 and this will be your y^. Then you can compare the y^ with actual y values.
@@navjotsingh8372 Thanks for the reply, are you saying the this particular sigmoid fn is the one that we will apply in the output layer ?
Yes, in the output layer, you have a classification problem. So you either need 1 or 0 for classification. Using sigmoid function as the output will help you get the output between 0 and 1. For the input layer, we use relu function so that we can avoid gradient vanishing.
Hello sir, ineuron link is no longer valid now. Please let us know from where we can download the material for these classes.
PURE JEM 💎
In 1:08:00 when you say we can use relu for hidden layers then won't we have the same problem of vanishing gradient again here ?
What I don't understand is why you're only using the neurons in backprop.
Like yea your weight is dependent on the prev neuron, but isn't that neuron depending on the weight and bias before it? Aren't you skipping a step by going neuron to neuron?
I think it should be 1/N not 1/2 in mean squares error cost function
Sir , I can't get the material of this lecture because the link is expired please have a look on my problem.
This content is amazing sir ❤️
i am unable to enroll for free in order to get the resources from dashboard and it showing me page not found error. Can anyone help me in resolving this please
good
content
SIr please provide the notess
Where i download activation jupyter notebook
where I can get the notes?? the link to the dashboard in the description is not working
I am unable to find material in ineuron portal now.. could you please check and update
@krishnaik sir please provide notes
what is the duration of Fullstack data science job guaranteed program.
1 year
This is super
How to got the notes in deep learning
why we need a Zero centered Activation?????
how do I get study material?
Ammazing sir ,LOVEDD it , just wow ... hinge innu jasti videos madi...dont stop the good work.
Did anybody got the resources??... I didn't able to find the resources in the dashboard...there are only video of day1 nd day2 class I got
please check the description and enroll int he community dashboard
@@krishnaikhindi Thank you for replying sir!!!.... Sir I have already enrolled
10/10
Hi Krish,
Can you please respond to email regarding FSDS course
Also i think the example calculation explained for soft Max. it shoul be like (e(net1))/(e(net1)+e(net2)+e(net3)) not like e10/e10+e20+e30 ...correct me if I'm wrong
fr mja aagya
can anybody share dashboard link ?
Check in the description of the video
you always using derivative, can you please tell me what is derivative here, it very very important i always confused
9/10