What is Recurrent Neural Network (RNN)? Deep Learning Tutorial 33 (Tensorflow, Keras & Python)

Рет қаралды 284,138

Күн бұрын

RNN or Recurrent Neural Network are also known as sequence models that are used mainly in the field of natural language processing as well as some other areas such as speech to text translation, video activity monitoring, etc. In this video we will understand the intuition behind RNN and see how RNN's work.
Deep learning playlist: • Deep Learning With Ten...
Machine learning playlist: kzfaq.info?list...
#recurrentneuralnetwork #rnn #rnndeeplearning #whatisrnn #deeplearningtutorial #rnnneuralnetwork #rnntutorial
Do you want to learn technology from me? Check codebasics.io/?... for my affordable video courses.
🌎 My Website For Video Courses: codebasics.io/?...
Need help building software or data analytics and AI solutions? My company www.atliq.com/ can help. Click on the Contact button on that website.
#️⃣ Social Media #️⃣
🔗 Discord: / discord
📸 Dhaval's Personal Instagram: / dhavalsays
📸 Instagram: / codebasicshub
🔊 Facebook: / codebasicshub
📱 Twitter: / codebasicshub
📝 Linkedin: / codebasics
❗❗ DISCLAIMER: All opinions expressed in this video are of my own and not that of my employers'.

Пікірлер: 140

@codebasics 2 жыл бұрын

Do you want to learn technology from me? Check codebasics.io/ for my affordable video courses

@jayshreeram6684 28 күн бұрын

I don't want to go in NLP can i avoid this ahead videos, because I have to focus on computer vision.

@sidsarkar8 2 жыл бұрын

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!

@architnarang3232 2 жыл бұрын

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

@shresthaditya2950 Жыл бұрын

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

@lindaeitelberg8841 2 жыл бұрын

Your explanation on recurrent neural networks is the best I've seen to date.

@buh357 2 жыл бұрын

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.

@hamzaashfaque2977 3 жыл бұрын

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.

@codebasics 2 жыл бұрын

Glad you like them!

@nurbekss2729 2 жыл бұрын

dude,where you have been all this time?!Great video!! Finally got a pure and exact explanation of ML models!!

@andreonofre2625 10 ай бұрын

You are a great teacher! Very easy to understand explanation about RNN!

@maitreyamoharil4299 8 ай бұрын

Best video i have seen till now about basic understanding of RNN Thank you

@crazysilvershine 2 жыл бұрын

Very crisp and to the point.Excellent explanation

@faddsfadds5690 11 күн бұрын

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.

@joycechepkemoichepkwony3530 2 жыл бұрын

I am still new in this. Trying to follow up all your videos. You've got good explanation. Thank you.

@codebasics 2 жыл бұрын

👍👍☺️

@ranjaninatarajasharma5917 2 жыл бұрын

Thanks a million for all super simple videos. God bless!

@ishaanrajpal273 Жыл бұрын

Your way to teaching is amazing!

@dlmsami2539 2 жыл бұрын

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

@seanisthegood Жыл бұрын

Seconded! What a guy!

@rayaneaboud9043 3 ай бұрын

great job taking your time explaining those notions thank you!

@Want_to_escape 3 жыл бұрын

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.😃👍

@codebasics 3 жыл бұрын

Most welcome!

@kmnm9463 2 жыл бұрын

Hi Dhaval ji, This is just amazing! Your insight into concepts of RNN and all deep learning is unparalleled. Just great. Thanks from Krish

@RounakPolley 3 жыл бұрын

This is probably the best channel for students learning AI Thanks a lot

@codebasics 3 жыл бұрын

Glad it was helpful!

@naffiahanger9316 2 жыл бұрын

Really awesome and simple explanation of RNN

@arjunreddy7203 3 жыл бұрын

Keep these coming! Loving them.

@codebasics 3 жыл бұрын

😊👍

@chocky_1874 2 жыл бұрын

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

@raselkarim2731 2 жыл бұрын

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.

@codebasics 2 жыл бұрын

Glad you liked it!

@zanwardrrknec953 2 жыл бұрын

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.

@codebasics 2 жыл бұрын

Thanks zanwardr for your feedback

@mohammadelghandour1614 Жыл бұрын

Thanks for the easy and simple explanation ! I just have a question: around 3:49 and about the issue #2, you were talking about the one hot-encoding issue that results in high computation cost in MLP. How does RNN solve this issue? and how are words represented in RNN if not in hot-encoded binary format?

@darshilvshah 3 жыл бұрын

Superb video !!!....in fact most of the videos of this channel are amazing !!!!

@codebasics 3 жыл бұрын

Glad you like them!

@texwiller7577 9 ай бұрын

Very very cool video! I discovered right now your channel and it seems very promising

@prasannan-robots 3 жыл бұрын

I was waiting for this.

@sain5275 Жыл бұрын

Happy Teachers day.. u r doing a great job .. 👏

@prabhamelady1680 2 жыл бұрын

Thanks for amazing Video with simple explanation :)

@jimnason7293 Жыл бұрын

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 Жыл бұрын

👍🏼👍🏼👍🏼

@jimnason7293 Жыл бұрын

@@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 :-)

@radionnazmiev546 6 ай бұрын

Simply you're the best man hands down!!!

@codebasics 6 ай бұрын

Thank you so much for your kind words.

@sejalgulati5286 3 жыл бұрын

So well explained. Thank you!

@codebasics 3 жыл бұрын

Glad it was helpful!

@shankar0015 3 жыл бұрын

Outstanding, keep the good work. thanks for the video.

@codebasics 3 жыл бұрын

Glad it was helpful!

@piyalikarmakar5979 2 жыл бұрын

Thank you sir.. I just have one query..we can use Feed forward NN for sentiment analysis and there also sequence of the words matters.. then how does that NN deal with that?

@Breaking_Bold 5 ай бұрын

Nice explanation of RNN !!!

@GaetanoCarmeloLaDelfa 22 күн бұрын

Amazing!

@buckybarnes9135 2 жыл бұрын

you explain so crystal clear , not on the face of it.

@RAZZKIRAN 3 жыл бұрын

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?

@codebasics 3 жыл бұрын

Raju, point noted. I will be adding all those videos that you suggested.

@strongsyedaa7378 2 жыл бұрын

@@codebasics How to do RNN with Sequential layers

@mastercomputersciencesubje3181 2 жыл бұрын

Wonderful explanation...made it very simple

@codebasics 2 жыл бұрын

Glad it helped

@zepingluo694 2 жыл бұрын

Thank you so much for the amazing videos!!

@codebasics 2 жыл бұрын

Glad you like them!

@rohithmandya2872 2 жыл бұрын

Nicely explained. I came back online just to give a thumbs up and subscribe. lol.

@himansushaw4938 Жыл бұрын

Nicely Explained

@izharkhankhattak 3 жыл бұрын

Good job. Thank you, Sir!

@codebasics 3 жыл бұрын

👍😊

@shefalisingh4675 Жыл бұрын

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??

@debatradas9268 2 жыл бұрын

thank you so so so much, Sir

@hardikvegad3508 3 жыл бұрын

FINALLY........😍😍😍 waiting for LSTM too...

@codebasics 3 жыл бұрын

yes LSTM will come up soon

@amanrai5285 Жыл бұрын

Thank You!!!!

@sagihaviv5675 3 жыл бұрын

sir how can i apply rnn for OCR? i want to build code recognition from image ? any tips and advice? im new to AI

@ShradhanandVerma Жыл бұрын

Nine explanation sir...

@narendrasompalli5536 2 жыл бұрын

Sir, do we must to add weights and bias?

@dsreebus 2 жыл бұрын

Well explained, better than tensorflow zero to hero

@samratsinghrathore7240 3 жыл бұрын

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

@parwaagrawal Жыл бұрын

Great Explaination.

@codebasics Жыл бұрын

Glad it was helpful!

@rajpurohitkiran5560 2 жыл бұрын

Sir, you have used hidden layer at different states. I had a doubt does the size of hidden layer remain same at each state

@cse8591 2 жыл бұрын

Great Explanation

@codebasics 2 жыл бұрын

Glad it was helpful!

@r21061991 3 жыл бұрын

Please make more videos on NLP !!

@SAURABHGUPTA_DINOSAUR 9 ай бұрын

Sir i have tried everything on the internet but my tensorflow isnt detecting my gpu and runs on cpu instead. What do i do?

@kmnm9463 2 жыл бұрын

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

@wellingtonereh3423 2 жыл бұрын

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

@sachinpriya88 2 жыл бұрын

Is Elman Recurrent NN or Simple Recurrent NN anonyms to each other or they r different in theory?

@prajwalbk5725 3 жыл бұрын

May the Force be with you and Grogu. Waiting for Friday's Bad Batch Ep 3.

@akshaytandon4792 2 жыл бұрын

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.

@sreyoshimitra6501 2 жыл бұрын

converting text sentences into vectors by Bag of words technique.

@ranishkalakshan3511 3 жыл бұрын

I am waiting for.it..please do a video about LSTM soon.

@codebasics 3 жыл бұрын

yup LSTM will come soon

@shanmugamnathan9915 2 жыл бұрын

For tumour detection RNN is suitable sir?

@vinaychitturi5183 3 жыл бұрын

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.

@strongsyedaa7378 2 жыл бұрын

codebasics Please answer this question

@starless7972 10 ай бұрын

gawd work

@shravannimma630 2 жыл бұрын

great way of explaining things ☺you should be teaching at university

@mugambo5505 Ай бұрын

nice clarification

@AKSHAY99552 3 жыл бұрын

Sir, can you please make a video on word2vec technique 🙏

@hp6hp1 3 жыл бұрын

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.

@codebasics 3 жыл бұрын

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/

@hp6hp1 3 жыл бұрын

@@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?

@nandakishorejoshi3487 2 жыл бұрын

@@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

@maheswarareddygajjala625 4 ай бұрын

if possible can you give us notes on this series

@user-jo2gx8ry6y 2 жыл бұрын

When you explain classifying is transaction fraud or not Sequence wasn't important but I think in real process in ANN input node has also label right? then if the input datas' sequence is different from before Consequently layer's weights were also changed. ( ex) input [1,2,3] and [3,2,1] bring about different output in ANN too. so I could say this model has sequence too!) Am I wrongly understand ANN? plz help me THANKS :)

@VinVin21969 2 жыл бұрын

yeah, this guy make a mistake . changing the sequence for ANN also give different output since changing sequence mean you put different vector value to each node which has different bias and weight .the thing is Neural network is also can used for NLP as long as it wasnt a complex sequence , and also even simple RNN cant be used for complex sequence too.. usually we using RNN-LSTM for complex sequence. hope this answer help you

@user-jo2gx8ry6y 2 жыл бұрын

@@VinVin21969 thx :)

@asifgul4810 3 жыл бұрын

Sir can u running an institute I want to learn data scientist I have done my graduation in commerce if no Then give me a one suggestion which institute is best for to learn data scientist

@amongus-rq2hb Жыл бұрын

You are How . As a indian it makes sense to me

@anshkumar-nw1qd 3 жыл бұрын

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)

@codebasics 3 жыл бұрын

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

@mreshma776 Жыл бұрын

Is rnn a neural network or it is an neural network algorithm

@subhashthangadipally4974 3 жыл бұрын

38 like & 300 viewer it's not my mistaken cz I didn't get any notification from KZfaq.

@parthpatwari3174 4 күн бұрын

"Tensorflow, Keras & Python" where are they in your video? although you explained the concepts nicely but please don't mislead people from next time

@dibsaha5593 3 жыл бұрын

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.

@codebasics 3 жыл бұрын

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.

@dibsaha5593 3 жыл бұрын

@@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

@Arjun147gtk 3 жыл бұрын

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.

@codebasics 3 жыл бұрын

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

@medhavanisharma3323 Жыл бұрын

Sir Can you share the Ha nd written notes..

@binilrajadhikari2643 3 жыл бұрын

sir, can you please explain the math needed for learning data structure and algorithm?

@ajinkyakadam4672 3 жыл бұрын

Playlist is not opening 😅

@codebasics 3 жыл бұрын

are you clicking on a link in video description? It opens fine for me

@raa__va4814 Жыл бұрын

im struggling to understand what multiple neurons are doing here ... shouldnt we just need one neuron to do all this

@shaiksuleman3191 3 жыл бұрын

Sir u r like doctor with out pain u give injection which will clear our confusion.

@codebasics 3 жыл бұрын

ha ha.. thanks shaik for your very poetic, analogy based appreciation and feedback. It is always a pleasure to read your comments my friend :)

@user-du3td6zw6d 2 жыл бұрын

Where is slides? i need slides.

@abcd-qq9zb 2 жыл бұрын

07:46:00

@basmahalsohbani5986 3 жыл бұрын

Why arabic translat can not

@vaizerdgrey Жыл бұрын

wait, is your name Dhaval?

@swaralipibose9731 3 жыл бұрын

Audio classification please 🙏

@codebasics 3 жыл бұрын

yes that will come up as well

@ansarbashashaik4345 3 жыл бұрын

@@codebasics thank you sir waiting for that

@chintapallitejesh8172 3 жыл бұрын

Bitcoin mining with Actuall Bitcoin ecosystem please

@codebasics 3 жыл бұрын

okay point noted

@manvirsingh3727 3 жыл бұрын

Please sir make a video on nlp

@codebasics 3 жыл бұрын

A dedicated NLP tutorial series is coming up soon

@manthanpanchal4071 3 жыл бұрын

Sir, I am 2020 pass out from the electronics branch, I am passionate about data science and want to do my masters abroad in data science, to be sure i have already gone through a data science boot camp, i my decision right ????????

@work-dw2hl 3 жыл бұрын

which country r u planning