RNN and Sequence model uses: 1. Completing Sentences 2. Translating Sentences 3. Entity Recognition 4. Sentiment Analysis Sequence is important here N.N work on numbers not strings so you will have to apply Hot-Encoding
this is brilliant, your explanation truly helps one build a genuine intuition for concepts. found your channel trying to understand whether USE or BERT would be better for my use case, so happy that I did
Your videos are so simple and easy to understand. They get across the basic intuition of Algorithms with ease. It's been challenging for a me to get in to this field and this content has been really helpful for me.
Thanks, I finally understand RNN. So basically you first encode each words in your sentence then feed encoded vector with weight to the neural network with activation function. Afterwards you feed next encoded vector with weight feed into neural network until feed all the words. So it’s like a loop, we only using on neural network. That a important point. And then we calculate the loss and use back propagation to adjust the weight and bias in our network.
When I do NLP processing now, I apply a Lisp or Prolog based approach where one matches everything at once. Years ago I learned that to understand a sentence, one must know the words, and to know the words, one must know the sentence. So the approach is to look at the whole sentence, and process the words and the sentence at the same time and as each part is discovered or combined, they reinforce the others. Nice video ----
@@codebasics its the ole 22 phrase, one cannot get a job without experience, so how do I get experience without a job" ----- you process both simultaneously from front to back, back to front, and the middle in between, as each word is identified, the phrase becomes identified, yet the phrase cannot be identified until the word is identified" Do multiple threading on the sentence at the same time, while continually combining into the largest sections and to the most discrete sections and bubble up and bubble down - bottom up, top down, meet somewhere in the middle where all parts agree ----------- so simple in an analogy ----- so now I code it once again :-)
Amazing. Sir, you may not be aware how easy you are making it. How close you are taking us to application. You must be taking huge efforts to make it so.
This is my firstv video in ai or deep leanring, but clearly understood, tq for the clear cut explaination, by seeing your video i am very curious to explore this field of ai and deep learning, thank you.
Hy bro, u r really awsmm. I don't really understand the class, explained by my professor. before going to the class, i will go through ur tutorials. it really helps me to understand well. thank you bro
Your videos are very informative and simple, but one thing you mentioned RNNs are sequence models but translation would not be always sequential when we translate an English sentence to French or Spanish, words may be here and there then RNNs would not work as efficiently. Is there any other model we can use??
thank u so much, expecting next topic videos in sequence, like LSTM,Bi-LSTM,GAN, please do videos with one example classification task like hate speech detection with these concepts like RNN, LSTM?
Hi, I have a doubt at at timestamp 09:00 in the video. In the NER example, for the input vector which one-hot encoding, for each word like Dhaval loves baby - there are 1's at two different places in the vector. Being one-hot encoded there should be only one '1' in each of the vectors right? This 1 corresponding to where the word appears in the vocab. Please clarify? Krish
Hello. The ground truth 1 0 1 1 is the y(answer). The network input is a word vocabulary( could be hot encoded or word embedding ). Each word has your unique vector
Finally I found you very luckily. Please resolve my query. I am newbie in python. So I want to finish only python basic then I want to move ML and want to complete all ML course with Python. Should I move to with Python basic only? I believe I will get satisfaction ans from you only.
sure. you should follow my first 15 videos in my python tutorials playlist and then first 9 videos in pandas playlist. After that you can directly jump to my machine learning playlist.
@@codebasics i have follow you always and i remember you are a very good guidance for us. Really this is great and valid information for me. so first 15 videos for python and first 9 videos for pandas this is fine to move your machine learning course. I mean no need anything to without this for begging to expert ML learn....... please let me
Sir! Most of the time, I have seen We predicting only number values from our Machine Learning Model. But I want know that How to find Category Values by taking Number values. Such as If I have a model that can predict a list of shop that giving Min price X and max price Y.
You asked an interesting question Himalay. We need to think about few things (1) is your base for output the current price meaning based on current prices from different shops you want to input a product and find shops between price x and y? That wouldn't be a machine learning problem, you can simply to price comparison using SQL query (2) Say based on some features we are trying to predict future price from different shops, in this case one option is to build separate model for each shop and train them. When you are ready for prediction you get predicted output for each shop and now do your price comparison for min x and min y. There is also multioutput regression which can predict multiple outputs, need to see exact problem statement, features to see if that can be utilized: machinelearningmastery.com/multi-output-regression-models-with-python/
@@codebasics Sorry, Sir. But I think you couldn't understand. I checked out your Real Estate project of Banglore city. It is giving prediction prices based on a few categories such as location, BHK, Bath, etc right? But If I want to location such as Akshay Nagar etc by giving Area, My budget as Estimate price and more. So I want to predict string Category, not number or int. Then what can do about it?
@@hp6hp1 , Neural network do not predict string as output directly. But they do predict word vectors or give probability distribution as output. We need to decode these output to get the respective output
Hi Sir, at 13.36, what will be the value of Y^(hat) at the first layer(Ironman)? Will it be 1 or [1 0 0 0] , asking this question as we are only passing oneword to the first layer.
See I request you to use analogy which everyone in India understands. In this video , the analogy is hindi words not understanding for us students of southern India. @codebasics you have fans throughout India but expecting respect for everyone equally
What does these functions in figure.canvas do? What is figure.canvas? fig.canvas.draw() fig.canvas.get_width_height() fig.canvas.tostring_rgb() What is fig.canvas if you could explain in detail. Also what does this code do: `image = np.frombuffer(fig.canvas.tostring_rgb())’ I tried hard but could not get any documentation related to them. Are they obsolete? Please help me I’m stuck
Sir was planning to do data analysis course according to your 3months plan wanted to know that i can do that course with any basic laptop of 4gb ram, i3 intel or i need much higher configuration laptop? (I'm asking about each and every tools which i have to learn)
It was pretty awesome, I have understood the RNN from your explanation and wondering about the word "Recurrent". And telling myself, ohhhh, that's why it's called Recurrent Neural Network.
you are a great man , thunk you for all what you did and what you are doing and what you wild do , you make deep learning looking like a simple math operation (1+1=2) , i encourage you to continue in this path of teaching
Sir, I need one advice. Have a job offer of data analyst, shall I take it and later move to data science and machine learning field or keep searching for machine learning related jobs? I wish I could have a word with you on chat/email.
Depends on your situation... are you in need of earning money immidiately if yes then take it. you can move to ML job later on as well but your experience as a data analyst might (or might not) create some hurdles for you in future. whereas if you start as a data scientist from your first job itself your resume would look clean
At 13:20 , it was mentioned to adjust weights after pass each sentence, while i remember in ANN the weights are adjusted after one epoch. Am I mistaken?
Hi. Thanks a lot for this explanation on RNNs. I had one question. At 4:25, you mention that converting the sentences to vector is called one hot encoding. Isn't it bag of word transformation? I request a clarification about this doubt. Thanks.
Thank you very much for this wonderful and simple explanation on RNN. I finally understood the concept. Lots of love from Saudi Arabia...Keep going. You are a genius.😃👍
Check out our premium machine learning course with 2 Industry projects: codebasics.io/courses/machine-learning-for-data-science-beginners-to-advanced
I don't want to go in NLP can i avoid this ahead videos, because I have to focus on computer vision.
Saw like 3-4 videos, no one really addressed that there is only one layer that is going through this process. By far the most clear video, thank you!
RNN and Sequence model uses:
1. Completing Sentences
2. Translating Sentences
3. Entity Recognition
4. Sentiment Analysis
Sequence is important here
N.N work on numbers not strings so you will have to apply Hot-Encoding
this is brilliant, your explanation truly helps one build a genuine intuition for concepts. found your channel trying to understand whether USE or BERT would be better for my use case, so happy that I did
I am still new in this. Trying to follow up all your videos. You've got good explanation. Thank you.
👍👍☺️
Your videos are so simple and easy to understand. They get across the basic intuition of Algorithms with ease. It's been challenging for a me to get in to this field and this content has been really helpful for me.
Glad you like them!
Thanks, I finally understand RNN. So basically you first encode each words in your sentence then feed encoded vector with weight to the neural network with activation function. Afterwards you feed next encoded vector with weight feed into neural network until feed all the words. So it’s like a loop, we only using on neural network. That a important point. And then we calculate the loss and use back propagation to adjust the weight and bias in our network.
When I do NLP processing now, I apply a Lisp or Prolog based approach where one matches everything at once. Years ago I learned that to understand a sentence, one must know the words, and to know the words, one must know the sentence. So the approach is to look at the whole sentence, and process the words and the sentence at the same time and as each part is discovered or combined, they reinforce the others. Nice video ----
👍🏼👍🏼👍🏼
@@codebasics its the ole 22 phrase, one cannot get a job without experience, so how do I get experience without a job" ----- you process both simultaneously from front to back, back to front, and the middle in between, as each word is identified, the phrase becomes identified, yet the phrase cannot be identified until the word is identified" Do multiple threading on the sentence at the same time, while continually combining into the largest sections and to the most discrete sections and bubble up and bubble down - bottom up, top down, meet somewhere in the middle where all parts agree ----------- so simple in an analogy ----- so now I code it once again :-)
Amazing. Sir, you may not be aware how easy you are making it. How close you are taking us to application. You must be taking huge efforts to make it so.
Thanks zanwardr for your feedback
This is my firstv video in ai or deep leanring, but clearly understood, tq for the clear cut explaination, by seeing your video i am very curious to explore this field of ai and deep learning, thank you.
dude,where you have been all this time?!Great video!! Finally got a pure and exact explanation of ML models!!
Keep these coming! Loving them.
😊👍
Best video i have seen till now about basic understanding of RNN
Thank you
I was waiting for this.
Superb video !!!....in fact most of the videos of this channel are amazing !!!!
Glad you like them!
Very crisp and to the point.Excellent explanation
Hy bro, u r really awsmm. I don't really understand the class, explained by my professor. before going to the class, i will go through ur tutorials. it really helps me to understand well. thank you bro
great job taking your time explaining those notions thank you!
Nicely explained. I came back online just to give a thumbs up and subscribe. lol.
Really awesome and simple explanation of RNN
Thanks a million for all super simple videos. God bless!
Your videos are very informative and simple, but one thing you mentioned RNNs are sequence models but translation would not be always sequential when we translate an English sentence to French or Spanish, words may be here and there then RNNs would not work as efficiently. Is there any other model we can use??
thank u so much, expecting next topic videos in sequence,
like LSTM,Bi-LSTM,GAN,
please do videos with one example classification task like hate speech detection with these concepts like RNN, LSTM?
Raju, point noted. I will be adding all those videos that you suggested.
@@codebasics
How to do RNN with Sequential layers
So well explained. Thank you!
Glad it was helpful!
Well explained, better than tensorflow zero to hero
which is best for forecating or prediction ... LSTM?
Hi,
I have a doubt at at timestamp 09:00 in the video. In the NER example, for the input vector which one-hot encoding, for each word like Dhaval loves baby - there are 1's at two different places in the vector. Being one-hot encoded there should be only one '1' in each of the vectors right? This 1 corresponding to where the word appears in the vocab. Please clarify? Krish
Hello. The ground truth 1 0 1 1 is the y(answer). The network input is a word vocabulary( could be hot encoded or word embedding ). Each word has your unique vector
Nice explanation of RNN !!!
Finally I found you very luckily. Please resolve my query. I am newbie in python. So I want to finish only python basic then I want to move ML and want to complete all ML course with Python. Should I move to with Python basic only? I believe I will get satisfaction ans from you only.
sure. you should follow my first 15 videos in my python tutorials playlist and then first 9 videos in pandas playlist. After that you can directly jump to my machine learning playlist.
@@codebasics i have follow you always and i remember you are a very good guidance for us. Really this is great and valid information for me. so first 15 videos for python and first 9 videos for pandas this is fine to move your machine learning course. I mean no need anything to without this for begging to expert ML learn....... please let me
Thanks for amazing Video with simple explanation :)
if possible can you give us notes on this series
Sir i have tried everything on the internet but my tensorflow isnt detecting my gpu and runs on cpu instead. What do i do?
Sir, you have used hidden layer at different states. I had a doubt does the size of hidden layer remain same at each state
Sir! Most of the time, I have seen We predicting only number values from our Machine Learning Model. But I want know that How to find Category Values by taking Number values. Such as If I have a model that can predict a list of shop that giving Min price X and max price Y.
You asked an interesting question Himalay. We need to think about few things (1) is your base for output the current price meaning based on current prices from different shops you want to input a product and find shops between price x and y? That wouldn't be a machine learning problem, you can simply to price comparison using SQL query (2) Say based on some features we are trying to predict future price from different shops, in this case one option is to build separate model for each shop and train them. When you are ready for prediction you get predicted output for each shop and now do your price comparison for min x and min y. There is also multioutput regression which can predict multiple outputs, need to see exact problem statement, features to see if that can be utilized: machinelearningmastery.com/multi-output-regression-models-with-python/
@@codebasics Sorry, Sir. But I think you couldn't understand. I checked out your Real Estate project of Banglore city. It is giving prediction prices based on a few categories such as location, BHK, Bath, etc right? But If I want to location such as Akshay Nagar etc by giving Area, My budget as Estimate price and more. So I want to predict string Category, not number or int. Then what can do about it?
@@hp6hp1 ,
Neural network do not predict string as output directly. But they do predict word vectors or give probability distribution as output.
We need to decode these output to get the respective output
Hi Sir, at 13.36, what will be the value of Y^(hat) at the first layer(Ironman)? Will it be 1 or [1 0 0 0] , asking this question as we are only passing oneword to the first layer.
codebasics
Please answer this question
Sir, do we must to add weights and bias?
You are How . As a indian it makes sense to me
Great Explaination.
Glad it was helpful!
Please make more videos on NLP !!
See I request you to use analogy which everyone in India understands. In this video , the analogy is hindi words not understanding for us students of southern India. @codebasics you have fans throughout India but expecting respect for everyone equally
38 like & 300 viewer it's not my mistaken cz I didn't get any notification from TH-cam.
Is Elman Recurrent NN or Simple Recurrent NN anonyms to each other or they r different in theory?
For tumour detection RNN is suitable sir?
great way of explaining things ☺you should be teaching at university
"Tensorflow, Keras & Python"
where are they in your video?
although you explained the concepts nicely but please don't mislead people from next time
sir how can i apply rnn for OCR? i want to build code recognition from image ? any tips and advice? im new to AI
What does these functions in figure.canvas do? What is figure.canvas?
fig.canvas.draw()
fig.canvas.get_width_height()
fig.canvas.tostring_rgb()
What is fig.canvas if you could explain in detail. Also what does this code do:
`image = np.frombuffer(fig.canvas.tostring_rgb())’
I tried hard but could not get any documentation related to them. Are they obsolete?
Please help me I’m stuck
Thank You!!!!
Sir Can you share the Ha
nd written notes..
Please sir make a video on nlp
A dedicated NLP tutorial series is coming up soon
Sir u r like doctor with out pain u give injection which will clear our confusion.
ha ha.. thanks shaik for your very poetic, analogy based appreciation and feedback. It is always a pleasure to read your comments my friend :)
Where is slides?
i need slides.
im struggling to understand what multiple neurons are doing here ... shouldnt we just need one neuron to do all this
nice clarification
gawd work
Sir was planning to do data analysis course according to your 3months plan wanted to know that i can do that course with any basic laptop of 4gb ram, i3 intel or i need much higher configuration laptop? (I'm asking about each and every tools which i have to learn)
I think it should be okay. I have not tried BI tools on such configuration so it is something you need to try an you will figure it out yourself
wait, is your name Dhaval?
Audio classification please 🙏
yes that will come up as well
@@codebasics thank you sir waiting for that
Chyna Key
Taylor Ruth Lopez Elizabeth Martin Shirley
Johnson Christopher Martin Jessica Garcia Sharon
Robinson Dorothy Lopez Carol Johnson Melissa
Moore Betty Gonzalez Patricia Thompson Jason
Martinez Betty Brown Jeffrey Allen Gary
Wilson Jennifer Moore Betty Clark Larry
Hall Cynthia Johnson Donna Clark Frank
Golgappa 😋😋
Taylor Michael Brown Michelle Wilson Barbara
Wilson Jennifer Wilson Paul Walker Michelle
Lee Laura Davis Anthony Brown Karen
i think we need AI to understand the speaker in this video...
Bitcoin mining with Actuall Bitcoin ecosystem please
okay point noted
Lewis Robert Jackson Robert Taylor Kevin
Moore Maria Rodriguez Nancy Young Donna
Wilson Edward Brown Nancy Lee Nancy
Young Jessica Gonzalez Shirley Jackson Brian
It was pretty awesome, I have understood the RNN from your explanation and wondering about the word "Recurrent".
And telling myself, ohhhh, that's why it's called Recurrent Neural Network.
Glad you liked it!
FINALLY........😍😍😍
waiting for LSTM too...
yes LSTM will come up soon
you are a great man , thunk you for all what you did and what you are doing and what you wild do , you make deep learning looking like a simple math operation (1+1=2) , i encourage you to continue in this path of teaching
Seconded! What a guy!
Sir, I need one advice.
Have a job offer of data analyst, shall I take it and later move to data science and machine learning field or keep searching for machine learning related jobs?
I wish I could have a word with you on chat/email.
Depends on your situation... are you in need of earning money immidiately if yes then take it. you can move to ML job later on as well but your experience as a data analyst might (or might not) create some hurdles for you in future. whereas if you start as a data scientist from your first job itself your resume would look clean
Allen Ruth Young Kimberly Jackson Donna
At 13:20 , it was mentioned to adjust weights after pass each sentence, while i remember in ANN the weights are adjusted after one epoch. Am I mistaken?
Hi. Thanks a lot for this explanation on RNNs. I had one question. At 4:25, you mention that converting the sentences to vector is called one hot encoding. Isn't it bag of word transformation? I request a clarification about this doubt. Thanks.
converting text sentences into vectors by Bag of words technique.
I am waiting for.it..please do a video about LSTM soon.
yup LSTM will come soon
Thank you very much for this wonderful and simple explanation on RNN. I finally understood the concept. Lots of love from Saudi Arabia...Keep going. You are a genius.😃👍
Most welcome!
Rodriguez Amy White Sarah Lopez David
Jackson Christopher Miller Ronald Harris William
Very good and clear. Thank you, Dhaval.
Johnson Mary Davis Angela Martinez Lisa
Miller Mary Rodriguez Donald Davis Scott
Walker Shirley Jackson Sharon Williams George
Allen Eric Robinson Larry Anderson Joseph
Garcia Edward Miller Kenneth Smith Michael
Jackson Carol Martinez Mary Smith Karen
Hall Anna Williams Jennifer Hernandez Sandra
Playlist is not opening 😅
are you clicking on a link in video description? It opens fine for me
Garcia Scott Smith Daniel Walker Barbara
Harris Jessica Johnson Helen Martinez William
Garcia William Martin Ronald Lee Paul
This is probably the best channel for students learning AI
Thanks a lot
Glad it was helpful!