God gifted teacher for us AI students. One in a million. Krish Naik sir also teaches, but nitish sir's explanation to the minutest detail so easily that it gets imprinted in the brain forever, no requirement of revising again. The topics are covered and explained so good that even a 12th pass student can also understand. May you live longer and keep teaching us this way.
yes it is 4, 1 but he transposed the matrix. The matrix is transposed so that you can take out the jacobian gradient. Which later will be used by gradient descent to get the answer.
Sir , one doubt in W1 when we are saying we have W1 metrics of 3 BY 3 and these are the weights , so when we are updating the W1 values using gardient descent , so are we modifieg those fillter values only . or that fillter values also have some diffrent W1 metrc which is actually modified . bvz if i am doing convolution with some fillter , say an edge detectoer fillter , then it is obvious that the value in that fillter should not chnge as by tht fillter i am trying to detect the edges .
@CampusX, should NOT the size of F remain (4, 1) even if we send in as mini-batch, as only 1 image will be passed in loop over the batch? Have you written son, as you combined them to avoid loops?
Sir pls make a video on k fold cross validation Sir maine abhut saara video dekha par usme acche se explain nahi Kiya gaya hai Aap ek video banao aur statistics par bhi video banao Apne playlist me dusre ka video daala hai Sir it's very important for me I am watching only your video , getting understand with your video only Humble request from honest student to best teacher
God gifted teacher for us AI students. One in a million. Krish Naik sir also teaches, but nitish sir's explanation to the minutest detail so easily that it gets imprinted in the brain forever, no requirement of revising again. The topics are covered and explained so good that even a 12th pass student can also understand. May you live longer and keep teaching us this way.
my god...why isn't this channel's subscribers beyond million...u are exceptional sir...sajda karta hun aapko sir jhukake!
One of the best explained playlists I've experienced so far. Hats off to teachers like you sir.
The way of explaining the topics is phenomenal.
Thanks
Thanks for your dedication to Indian AI society
finally i came to know how does backpropagation in CNN, Thanks a lot ..
I think I have to watch it again , bit complex😀
@@rajatkurhade1040 to dekh n usko ku bata r h 😂😂😂😂
r u single?
Enjoy your teaching style, all things to the point.
Thank you so much for beautifullly explaining such a hectic topic .
waah! sir kya samjaya hain! Data Science apse hi compele karege hum!
Thank you!
You’re best Nitish sir❤️
best explanation on internet
thank you so much sir from bottom of heart
you are god for me. like to banta h sir
Excellent. So simple and easy to understand. Keep it up Dear.
Amazing Explanation
Please make some videos on onnx conversion as well as you have good knowledge and also explaining in detail which helps easy to understand. Thanks
hats off to your efforts. :) iam watching daily
what a beautiful Explained sir
best playlist
Thanks a lot sir , Wanted a video on this topic 👍
great explanation!
Hi Nitish,pls cover learning rate scheduler,call backs,tensorboard operations
Best video sir
finished watching
Thank you sir🤩...for the amount of effort you put to make the complex topic very simpler
Killer explanation👌
Help a lot.
Thanks
Like and comment everyone to make this channel's reach wider and broader.
super clear - thank u so much for such great content sir..
crisp
but had to watch 3 times to grasp such a heavy load of info😅
kudos to u sir
really
Helps us a lot
Big help❤❤❤
Best Explained.
hii
Sir , please complete this series .
Sir !! Make 2-3 videos in a week 🙏
sir companies me humko khudse python codding karke karna hota machine learning stuff ya phir sklearn se kar sakte hai??
Hello sir please complete/upload more videos in the PYTHON series. Thankyou.
Why do we multiply the formula by 1/m
ig,the derivative(lhs) is for 1 image but the rhs is for m images so for lhs to be equal to rhs we multiply rhs by 1/m
35:17 - size of F you mentioned (m,4). Since F is a flattened array, shouldn't the size of F be (4m,1)
yes it is 4, 1 but he transposed the matrix. The matrix is transposed so that you can take out the jacobian gradient. Which later will be used by gradient descent to get the answer.
Great Video
Sir , one doubt
in W1 when we are saying we have W1 metrics of 3 BY 3 and these are the weights ,
so when we are updating the W1 values using gardient descent , so are we modifieg those fillter values only . or that fillter values also have some diffrent W1 metrc which is actually modified .
bvz if i am doing convolution with some fillter , say an edge detectoer fillter , then it is obvious that the value in that fillter should not chnge as by tht fillter i am trying to detect the edges .
Done!
how are we going to justify (1,m) shape of delL/delb2 in case of batch_size=m, shouldnt it be (1,1)
@CampusX, should NOT the size of F remain (4, 1) even if we send in as mini-batch, as only 1 image will be passed in loop over the batch? Have you written son, as you combined them to avoid loops?
but total images are m so (4,m)
Is cnn model trained for every pridiction of image classification? Like for every prdiction model needs to trained
for different test data model needs to be trained
only then we can get different outputs
Sir pls make a video on k fold cross validation
Sir maine abhut saara video dekha par usme acche se explain nahi Kiya gaya hai
Aap ek video banao aur statistics par bhi video banao
Apne playlist me dusre ka video daala hai
Sir it's very important for me I am watching only your video , getting understand with your video only
Humble request from honest student to best teacher
❤❤❤
Sir, Please start object detecion
11:35-->19:55Logical flow of our CNN architecture.
goat
sir one thing i didn't get how A2 in an matrix, as i know we did calculation of A2= [W2dotF + B2] and this is only single element
time=26:00?
respect
too good
why is his videos r so addicting ?
Chalo Aaj 2nd hi sahi😄
Gem
1/m se multiply kyon karna padega sir, agar 1,m ki image hai to?
💛
caa anyone please confirm me that after applying pooling of order 2 by 2 the obtained matrix is of order 3 by 3?
Hello sir, would request you if you can make a video specific to 'Dython' library
Thanks
so if cost function is quadratic loss, all of the equation will change na ?
🤟
Just open RUclips 😀 ki kahi sir ka video to nhi aa Gaya
Main first
i hope our comments/likes gives u motivation :)
sir part 2
Sir how to create rest API in python and java pr videos btaiye pllz
are bha pehle deep learning plalylist to complete ho jane de
Crack it 🎇
if ur videos are not in english, please put the language in the title
Bhai comments bhi ek baar dekh lo 😊
bro,try english,it helps many more people
Sir, please complete the series!!!