Deep Learning using Matlab

Поділитися
Вставка
  • Опубліковано 10 лис 2024

КОМЕНТАРІ • 176

  • @ZakWlak
    @ZakWlak Рік тому +1

    Let's explain in previous videos that the whole idea behind deep learning is the back propagation algorithm. So in this video where the viewers are expecting an explanation of the back propagation algorithm, let's write a bunch of lines and tell the viewers "I think you understood that this is the back propagation algorithm". Bravo.

  • @mohamedabdallah400
    @mohamedabdallah400 4 роки тому +4

    You are one of the greatest explainer of deep learning, thank you very much, gazakallahokhiran, from egypt

  • @merlissoto3923
    @merlissoto3923 3 роки тому +1

    Like and thank you very much for these tutorials!!!
    I have just finished this exercise and my neural network works perfectly... Thamk you!
    First of these three sections over.

  • @abdullahalmazrouei9044
    @abdullahalmazrouei9044 5 років тому +3

    For W1=2*rand(20, 25)-1 i think it should be W1=2*rand(25,20)-1. According to your source code the first parameter i.e. 25 represent the number of the previous layer, whereas the second parameter i.e, represents the number of nodes related to the next layer.

    • @skalar-haubitze1619
      @skalar-haubitze1619 4 роки тому

      It is correct since it is exactly the other way around: the first number represents the number of the next layer, whereas the second parameter represents the number of the previous layer. Also I believe that if you would switch these two parameters by mistake you would run into an error because of wrong matrix dimensions.

  • @AliAhmed-f9o1m
    @AliAhmed-f9o1m Рік тому

    Great. you are a high level lecturer. I have got more benefits from this video. Many thanks.

  • @zayyanushuaibu6188
    @zayyanushuaibu6188 5 років тому +5

    You are truly a genius, May God increase you in knowledge

  • @ahedalbadin2889
    @ahedalbadin2889 5 років тому +2

    Sir. You Are Amazing. Thank you very much. I know the NN theoretically but recently I needed to understand the Deep Learning process in few words.You've made my day.
    One Note: could you test the algorithm with not a perfect input and show the error.

  • @zuzanna6385
    @zuzanna6385 3 роки тому

    Thank you! You saved my life and project! Big love!!!!!

  • @imrankanj00
    @imrankanj00 6 років тому +1

    Your contents are well arranged. I have couple of questions. 1) what is the logic of exponential and division in softmax function? 2) In my case I have an image of 20x30 with output male/female, how many input and output nodes, I need?

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  6 років тому

      Thank you. You asked a good question.
      1) please plot the output of 'softmax' function separately. It will answer your question. It is difficult it express it properly using text.
      2) for every pixel, there will be one input nodes. For 20*30 image, you need 600 input nodes.
      3) There are two classes - (i) male and (ii) female. So two output nodes.

    • @imrankanj00
      @imrankanj00 6 років тому

      @@NuruzzamanFaruquis Thank you for your reply. Now I get it. Is it possible, you suggest some book/material where I can get a detailed and more complex example. I know this question is not related to this video but if you can help, I will appreciate. Thank you.

  • @3dp4e1
    @3dp4e1 2 роки тому +1

    Hi! Could you put a video about preparing a datastore for a multiple-input image classification CNN in Matlab? Thanks

  • @jeancarlosadrianza7416
    @jeancarlosadrianza7416 5 років тому

    Hi, Nuruzzaman. Your video "Neural Network using Matlab" doesn't show its comments (at least in my PC). For this reason, I want to congratulate you for that great video. Your explanations are very clear.
    I wanted to create a more general version of your SGD_method() to solve a exercise that I must to do, however, when I arrived to the calculation of delta and deltaweight, I realized that both calculations would produce incompatible multiplication of matrices. Please let me show you:
    delta = output * (1 - output) * error;
    If I treat output like a array, I don't know the procedure to calculate delta. In my opinion it's impossible, because if (for instance) output has a order 2x1, (1 - output) also would have this order. Then we will have a multiplication of 2x1 and 2x1... and then we have the array error.
    Something similar happens with deltaWeight.
    I would like to know if you can make a video to show how to handle the training and testing of a perceptron with more than one output. My bigger doubt is: how is affected the matrix of weights when a error appears in (for instance) the output 1? The error can affect every element of the matrix or only the weights related with that output?
    Best regards and congratulations again!

  • @kanwalnaz2305
    @kanwalnaz2305 3 роки тому +1

    Excellent work thanks

  • @NicoleQuimper
    @NicoleQuimper 4 роки тому +1

    thank you for these tutorials! the visuals are super helpful, there no way i could understand this without them you have an awesome teching method!!!

  • @mursalinkabir
    @mursalinkabir 6 років тому +2

    A solid guide to deep learning

  • @guido.demedici
    @guido.demedici 2 роки тому

    Congratulations to your really well made videos

  • @Elmusrati_Channel
    @Elmusrati_Channel 5 років тому +3

    Thank you for the well prepared video. But this is just the conventional backpropagation algorithm known since 30 years. I did the same using basic language more 23 years ago. It would be better if you show how to use deep learning ready functions in Matlab. No one need to write the details of coding nowadays. Thank you

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  5 років тому +1

      I agree with you.
      With such experience you do not need the details of coding. However, I have some enthusiastic new learners around. They want to learn from the very basics.
      Thank you for your valuable suggestion.

    • @zeyadalabsy7332
      @zeyadalabsy7332 3 роки тому

      @@NuruzzamanFaruquis I have a question, why do the weights become NAN if we set 1 to more than one value in the outputs? for example if the first output is [0 1 1 0 0], the weights become NAN. Could you please explain it?

  • @kumudaranimahanta7634
    @kumudaranimahanta7634 2 роки тому

    can you make a video explaining all the basic things like nnet, epoch, newp etc and code of training neural network in matlab

  • @muhibullah2134
    @muhibullah2134 4 роки тому

    thanks for sharing your valuable lectures

  • @xdgamestudios2660
    @xdgamestudios2660 6 років тому +3

    It's me from the other video, Thank you soo much for you tutorials and explanations, I think I'm starting to understand how I should do it!! one other thing... in the case I'm working I have 2000 input nodes and only a binary output... 1 or 0 how should the correct output be?! only one value?! I don't know I can I make it end with one single node!

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  6 років тому +2

      Take two output nodes - [1 0] for true and [0 1] for false. Otherwise even for no signal, there will be confusing false result.

  • @123nops
    @123nops 3 роки тому

    This video is very helpful. I am doing a cryptanalysis on my research right now using deep neural network. May I ask if I will use this code on your MATLAB demo, on which part should I define my 128 input nodes and 128 output nodes? Yours has 25 input and 5 output nodes? Thanks for the help.

  • @neelabhjyotisaharia1979
    @neelabhjyotisaharia1979 2 роки тому

    Was helpful...One question: How to back propagate when there are also biases with the weights?

  • @kavaskarsekar2835
    @kavaskarsekar2835 5 років тому +1

    Hi , I understand the DL concept but have a question that how to choose hidden layers and hidden neurons. Is there any specific rule or trial and error?

  • @ahmadkelixo7243
    @ahmadkelixo7243 3 роки тому

    permission to learn sir. thanks you. from Indonesia

  • @LoYalEditZ07
    @LoYalEditZ07 6 років тому +2

    Bests of the best. Great video

  • @fatihaji5339
    @fatihaji5339 3 роки тому

    Hi, the video was very good. To set up the network, my input data is 1000 x 3 and my target data is 1000 x 1. The output of training and testing their values is more nan. Why is this happening? What are the reasons for this?

  • @irfanuddin1948
    @irfanuddin1948 6 років тому +1

    indeed a good tutorial to have an idea about deep learning

  • @ibrahimhaddadi2812
    @ibrahimhaddadi2812 6 років тому +1

    this is a great point to start . thank you so much ,

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  6 років тому

      You are welcome. Feel free to ask if you face any problem.

  • @willdonell8508
    @willdonell8508 4 роки тому +2

    Hello,
    I have one question. Is your delta corresponding to the generalized delta rule?
    I can’t find the derivative of ReLU function
    Good tutorial!! It’s very helpful for me👍
    Thank you very much for sharing

    • @AkashSingh-rv8vx
      @AkashSingh-rv8vx Рік тому

      Derivative is 1 in non negative region. Elsewhere zero. Not defined at zero though.

  • @snehalgaikwad6436
    @snehalgaikwad6436 4 роки тому +1

    Plz take lecture on hopfield neural network to train the image

  • @harikrishnaponnam4680
    @harikrishnaponnam4680 4 роки тому

    Thank you very much it's very helpful for me.

  • @AndreaBarucci80
    @AndreaBarucci80 5 років тому

    Thank you vert much, well done and very useful! I hope you will insert other videos about Deep Learning and Machine Learning using Matlab.

  • @abdullahullash8471
    @abdullahullash8471 6 років тому +1

    very informative. looking forward for more videos.

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  6 років тому

      Thank you. I hope this lesson will help you to understand the concept of DNN and train DNN easily.

  • @durgasreeraghavan2049
    @durgasreeraghavan2049 4 роки тому +1

    I am chemical engineer, I am new to this machine learning programs. My thesis involves ANN. I have the same question as natalia chu has. How do I decide no. of neurons in hidden layer?

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  4 роки тому

      I am still studying about it. I will share my findings in upcoming lessons.

  • @raihanrafique7406
    @raihanrafique7406 4 роки тому

    Great Presentation,

  • @عبدالوهابفوزى-ث8ط
    @عبدالوهابفوزى-ث8ط 3 роки тому

    Thank you for these tutorials!

  • @nuttaponkasemsuk7670
    @nuttaponkasemsuk7670 5 років тому

    Did you use sigmiod to calculate the weight as taught in the clip first?

  • @know_edu1406
    @know_edu1406 4 роки тому

    Fine, your explanation. Can you tell in which ways (CNN or, Sparse Autoencoders or RBMs or LSTM are used to implement this Deep Learning?

  • @Akshay-cl5mz
    @Akshay-cl5mz 3 роки тому

    Sir, can i use this method for estimation of heat for my batch reactor temperature control ? Please respond

  • @ferdaozdemir
    @ferdaozdemir 5 років тому

    Hello Again, I ran the example and updated it for a larger input set. The updated version works but the softmax function returns Nan after a few samples during training which affects the whole training session. I searched for a safe version of softmax but couldn't find one. I found one example with phyton np. I install phyton and implemented the function but getting errors such as "Conversion of MATLAB 'sym' to Python is not supported". I would be appreciated if you can suggest a safe softmax function implementation?

  • @MuhammedAliErbir
    @MuhammedAliErbir 6 років тому +1

    teşekkür ederim kardeşim, anlaşılır şekilde anlatmışsın, eline sağlık
    Kırıkkale Üniv. Bilgisayar Müh. Doktora Talebesi
    Muhammed Ali ERBIR

    • @basharsaad4142
      @basharsaad4142 5 років тому

      i'm in cukurova university ,Turkey , ADANA are you working on image processing like object detection

  • @kamgangblaise2645
    @kamgangblaise2645 3 роки тому +1

    Thank you for the video. I have a question. What if your inputs and output to the network are sensors data, how can it be done because i am having some troubles

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  3 роки тому

      You need a data processing module in between the sensor node and the input nodes of the network. That's all you need to do.

    • @kamgangblaise2645
      @kamgangblaise2645 3 роки тому

      @@NuruzzamanFaruquis Data processing module?

    • @kamgangblaise2645
      @kamgangblaise2645 3 роки тому

      @@NuruzzamanFaruquis I have 3 inputs data (1 reference and 2 sensor data) and 1 output data

  • @nataliachu2495
    @nataliachu2495 5 років тому +5

    Thank You for lecture! I am still puzzled - why exactly 20 neurons in hidden layer? Could You, please, provide formula for calculation?

    • @rajasekarsuriya6269
      @rajasekarsuriya6269 4 роки тому

      Bro,y he is using only 20 nodes in hidden layer.Y is it so??

    • @mathi_s
      @mathi_s 2 роки тому

      @@rajasekarsuriya6269 Hi, did you got the answer for how to optimize the number of neurons.

  • @merveozdas1193
    @merveozdas1193 2 роки тому

    Thank you so much for your sharings :) 👏👏 but I have a question, if I use correct_output(:,:,k)', is it true or if I use input_image(k,:)', is it possible? I couldn't understand the matrices what you mean

  • @ofijanabu4689
    @ofijanabu4689 4 роки тому

    thank you very much Ifound it as best lecture would you please tell me how to realize a communication system on deep learning?

  • @BappaMukherjeeismdhanbad
    @BappaMukherjeeismdhanbad 3 роки тому

    can you please make a video on how to fix the optimum number of layer and what should be the neurons in these layers

  • @AkashSingh-rv8vx
    @AkashSingh-rv8vx Рік тому

    Hello Nuruzzaman Sir, can you explain how you arrived at error of hidden layers as the product of next weight and delta of next node?
    Please share the source if you are short of time.

  • @KraiemRayene
    @KraiemRayene 5 місяців тому

    Hi, thank you for this video. can you send me the datasets your used for this application

  • @arcimran5553
    @arcimran5553 6 років тому +1

    Knowledgable....& informative topics..... carry on.

  • @chetansoni2795
    @chetansoni2795 4 роки тому

    The number of hidden neurons should be 2/3 the size of the input layer, plus the size of the output layer

  • @jonathancohen64
    @jonathancohen64 4 роки тому

    How can I work out which input is the most effective / had the largest meaning to the output of a neural network

  • @ashik2052
    @ashik2052 6 років тому +2

    Informative
    Carry on genius
    Waiting for ur nxt Video

  • @supriyanaik316
    @supriyanaik316 4 роки тому +1

    Can I give any 2d cad geometry data as an input to deeplearning in matlab??

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  4 роки тому

      Yes, you can. You have to process your data accordingly.

    • @supriyanaik316
      @supriyanaik316 4 роки тому

      @@NuruzzamanFaruquis can you give some ideas that how to start it?? Should I convert all 2d cad geometry into image file or what???

  • @md.ashikurrahmankhan582
    @md.ashikurrahmankhan582 4 роки тому +1

    Thank you for you nice presentation. However, during running the code with Matlab2016a I found error as follow
    function [w1, w2, w3, w4] = DeepLearning(w1, w2, w3, w4, input_Image, correct_Output)

    Error: Function definitions are not permitted in this context.

    function y = ReLU(x)

    Error: Function definitions are not permitted in this context.

    Undefined function or variable 'ReLU'.
    >> TrainingNetwork
    Undefined function or variable 'ReLU'.
    Error in DeepLearning (line 9)
    output_of_hidden_layer1 = ReLU(input_of_hidden_layer1);
    Error in TrainingNetwork (line 46)
    [w1, w2, w3, w4] = DeepLearning(w1, w2, w3, w4, input_Image, correct_Output);
    Could you suggest where is the problem! Thanks in advance.

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  4 роки тому

      I have rechecked the code. It is working properly in my end. Would you please switch to MATLAB 2018B or layer version and try again?

  • @raghad3189
    @raghad3189 3 роки тому

    Thank you very much , but i have quastion when i write save ('neoral network.mat') and load('neoral network.mat')
    Matlap saing this error !why??

  • @hakanbaki509
    @hakanbaki509 4 роки тому

    hello,
    i have a question i have implemented the neccessary argumets to create XOR funcion with the code i have learned just by now. however when i apply [0;0] as it descibes the very first rule of XOR gate, i am getting 0.2500 in each segments while it must give me apprximately 1 in the first input. why i am facing that problem? edit: well i might saw the problem. just because all the inputs are 0 at the first rule of gates, none of this gate implementations will fit to your code due to the hidden layer values will be zero(for all output it is getting the equal sign which is 0.25). so i need to think about something else to implement gate configuration of ANN.

  • @ahmedbaba3715
    @ahmedbaba3715 5 років тому +1

    Thanks for the explanation ... Please, I want a code (neural networks) in order to check the quality of water ..... and God bless you good

  • @lama8oct
    @lama8oct 3 роки тому

    I wrote the code like you did in the video but i had error, how can i solve it??

  • @vamsiakula653
    @vamsiakula653 4 роки тому +1

    Why are they 20 nodes in the hidden layer?
    Please answer

  • @abdullahalmazrouei9044
    @abdullahalmazrouei9044 5 років тому +2

    Goodmorning sir,
    Why you used exactly 20 nodes for hidden layers? I did research on books but i hzvennt understand how to choose the correct number of nodes for hidden layers.
    Kind regards Doctor.

  • @chetansoni8620
    @chetansoni8620 4 роки тому

    sir can i use this same code.. instead of our digit (matrix) can we convert an RGB image to binary - then with binary matrix suppose the already known output i will give which is same as in this video... If i do this then how much epoch will i need if i will give 20 images of 28*28 pixels..?? please answer sir

  • @ahmadalghooneh2105
    @ahmadalghooneh2105 5 років тому

    May the elder gods bless you with even more knowledge, thank you thnk you thnk you

  • @themostimportantwatchtowat8943
    @themostimportantwatchtowat8943 5 років тому

    Thanks. It's a really great tutorial

  • @shihabo3
    @shihabo3 6 років тому

    well organized video

  • @tnpscmaterial627
    @tnpscmaterial627 4 роки тому

    Sir, how to recognition images in video scene using deep learning in matlab

  • @Tribeball359
    @Tribeball359 6 років тому +1

    Thank you so much for your lecture! Do you have any example on how to do semantic segmentation of an object in an image? And what if the input image is not a grayscale or binary? For example RGB or RGBD image?

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  6 років тому

      Right now I do not have any example on semantic segmentation. However, I will try to prepare one for you.
      And and RGB image has three different values for single pixel. The pixel values are the input signal. We cannot take three input signal at the same time in a single input node. That's why we use grayscale image.

    • @Tribeball359
      @Tribeball359 6 років тому

      @@NuruzzamanFaruquis Thanks a lot! So, for multi channel input we need to modify the DCNN structure...

    • @meghnatripathi5571
      @meghnatripathi5571 5 років тому

      @@NuruzzamanFaruquis sir can you please help me on a code to find wear rate for my project . i am really stuck???

  • @SABIRsabir-tn2oj
    @SABIRsabir-tn2oj 4 роки тому

    thank you its very good idea

  • @merlissoto3923
    @merlissoto3923 3 роки тому

    Teacher, there is something I don´t understand.
    In the section of adjustment at the DeepLearning function:
    adjustment_of_w4 = alpha*delta*output_of_hidden_layer3';
    adjustment_of_w3 = alpha*delta3*output_of_hidden_layer2';
    adjustment_of_w2 = alpha*delta2*output_of_hidden_layer1';
    adjustment_of_w1 = alpha*delta1*reshaped_input_Image';
    alpha is a real number, delta is a (5x5) matrix multiplied for (output_of_hidden_layer3') wich is a (1x20) matrix... I don´t get this part
    And similar for the rest:
    delta 3 is a (20x5) matrix, and (output_of_hidden_layer2') is a (1x20) matrix
    delta 2 is a (20x5) matrix, and (output_of_hidden_layer1') is a (1x20) matrix
    delta1 is a (20x5) matrix, and (reshaped_input_Image';) is a (1x25) matrix
    How are this multiplication among matrix possible? and I know it works perfectly because I did the exercise
    Hope you might read and answer this message.
    Thanks in advance!
    Atte,
    Merlis.-

  • @niranjanshegokarniranjanshegok
    @niranjanshegokarniranjanshegok 3 роки тому

    Thanks for nice tutorial. Please provide the ReLU.m file.

  • @dr.b.srinivas5505
    @dr.b.srinivas5505 6 років тому +1

    Excellent.

  • @tkhankhoje39
    @tkhankhoje39 5 років тому

    can u upload for simple inputs and output values with normalization of variables?

  • @tonikpinik7392
    @tonikpinik7392 5 років тому

    Really it's amazing...

  • @hannahhumphreys3002
    @hannahhumphreys3002 6 років тому +1

    ataScience New York 2019(January 23-24)
    Join us in New York, January 23-24, for this unique peer-to-peer event for top data science minds. Explore how to drive innovation within your business through a data-driven culture, build a world-class team, and utilize the latest innovation to solve some of the toughest problems facing corporate America.
    Book your tickets: bit.ly/2OWuepR

  • @dragonex68
    @dragonex68 4 роки тому +1

    thanks for the videos. this is useful during lockdown haha

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  4 роки тому

      Glad you liked it!

    • @vamsiakula653
      @vamsiakula653 4 роки тому

      @@NuruzzamanFaruquis Why are they 20 nodes in the hidden layer?
      Please answer

  • @AmmarAhmedSiddiqui
    @AmmarAhmedSiddiqui 5 років тому

    what if we feeed an input image which is not of "1" but close to "1" ?

  • @cherylli4769
    @cherylli4769 5 років тому

    Very helpful, thank you!

  • @Profeamgad
    @Profeamgad 5 років тому

    Thanx How can I obtain equation model from nn-tool matalb

  • @LoYalEditZ07
    @LoYalEditZ07 6 років тому +1

    Can you explain neural networks algorithm with machine learning in prediction of anything

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  6 років тому

      Are you asking about predictive model using neural network?

    • @LoYalEditZ07
      @LoYalEditZ07 6 років тому +1

      @@NuruzzamanFaruquis yes. In Matlab plz with code also. I'm beginner in learning all these concepts.

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  6 років тому +1

      @@LoYalEditZ07 I understand. I will try my best to prepare a tutorial that will help you building solid understanding.

  • @basharsaad4142
    @basharsaad4142 5 років тому

    Thank you sir can i do this code for gray image ? ???

  • @dr.avijendar5631
    @dr.avijendar5631 5 років тому +2

    sir..can you provide code for image classification using DNN

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  5 років тому +1

      Sure I can. Would you please tell me which classification (binary or multi-class) you are talking about?

    • @dr.avijendar5631
      @dr.avijendar5631 5 років тому +1

      @@NuruzzamanFaruquis multi class mr image classification i need

    • @user-nf7iu2qg4b
      @user-nf7iu2qg4b 5 років тому

      Hello sir .. iam Appling phd .. I need your help please

  • @МихайлоДимянчук
    @МихайлоДимянчук 5 років тому

    Why you used exactly 20 nodes for hidden layers?

  • @anushkabhatt1129
    @anushkabhatt1129 4 роки тому

    I have a numeric matrix as input file for training. Can I train that using deep learning

  • @darmatv9882
    @darmatv9882 5 років тому

    Hi! I truly enjoy your video. I even tried to run the same code in MathLab R2016a, but it is given me an error message "Not enough input arguments reshaped_input_image = reshaped(input_image(:, :, k), 25, 1);
    " I want to use this method for face recognition, because I am conducting research on a topic Improving the effectiveness of Deep Learning Approach for Face Recognition . How can I see the designed Architecture of this Algorithm in the Matlab that is the Network Architectural design and how to input face images into the network?

    • @zayyanushuaibu6188
      @zayyanushuaibu6188 5 років тому +1

      you dont have to run this file that shows error, run the the test file straight it will work.
      run=deepnet then testdeepnet

  • @zainabhaddad3390
    @zainabhaddad3390 5 років тому

    hi , I need a help for preparing an RCNN multiclasse (multi object detection ) code in matlab can you help me please

  • @qusayhamad5741
    @qusayhamad5741 5 років тому

    thank you good work

  • @username42
    @username42 5 років тому

    any chances to get the codes and the files from the video ?
    btw, we dont test our model with the same data we trained, it does not make sense, it is totally biased. you should have been created another data set for testing. and what about the validation of the model you have been trained. so here u have just only train the model with just 5 input without any validaiton and then use the same 5 inputs as showing testing which is not at all neither can be also as validation.

  • @foodsscenes5891
    @foodsscenes5891 5 років тому

    Thanks for the useful instructions; I have a project which needs your help. How can I contact you?

  • @LYu-rw5ze
    @LYu-rw5ze 5 років тому +1

    very helpful, thank u!!

  • @agungsusilo530
    @agungsusilo530 5 років тому

    thank you so much sir, i tried exactly the sam ecode but it says that "Error using - Matrix dimensions must agree." can you help me with this problem? thank you in advance

  • @البداية-ذ1ذ
    @البداية-ذ1ذ 5 років тому

    I have question please my laptop is cpu and i want to ran this code with data consists of arround around million samples ,can i do that

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  4 роки тому

      Yes. You can do it. However, the training process will take longer.

  • @jawadur_
    @jawadur_ 6 років тому +1

    useful!

  • @valerianherzog6597
    @valerianherzog6597 4 роки тому +1

    Thank you for this very helpful videos! Could you explain how you get the number of neurons in the hidden layers? Is there a specific rule?

  • @SA-pg5jt
    @SA-pg5jt 2 роки тому

    thanks

  • @meghnatripathi5571
    @meghnatripathi5571 5 років тому

    why are there 20 hidden layers ?

  • @dodgeridersrt5650
    @dodgeridersrt5650 5 років тому

    How can apply this for colored image ?! Instead of grayscaled image ?

    • @thelegend5248
      @thelegend5248 5 років тому

      This is a binary image. I have the same question. How can it be applied to grayscale image?

    • @chetansoni8620
      @chetansoni8620 4 роки тому

      @@thelegend5248 I think by converting gray to binary and then by using binary matrix....

    • @thelegend5248
      @thelegend5248 4 роки тому

      @@chetansoni8620 This idea is also fine. But actually I want to train deep learning with colored image. How to do that

  • @RahulSharma-xo5nd
    @RahulSharma-xo5nd 4 роки тому

    Please make a video on medical image i.e. .nii file

  • @nitinsingh3018
    @nitinsingh3018 5 років тому

    sir i also need code for image classification using DNN. can you provide it.

    • @nitinsingh3018
      @nitinsingh3018 5 років тому

      sir my email id is nitin1759@gmail.com

  • @rordic.y5947
    @rordic.y5947 4 роки тому +2

    Did anyone take the homework of "why 20 notes"?

    • @vamsiakula653
      @vamsiakula653 4 роки тому

      Same question from my side? What is the answer?

    • @AkashSingh-rv8vx
      @AkashSingh-rv8vx Рік тому

      I think one can take any number of nodes from structural point of view which will only effect the number of weights. However, this selection will impact the performance of model.

  • @haram812
    @haram812 4 роки тому

    adjustment_of_w4 = alpha * delta * output_of_hidden_layer3;
    all steps are correct but i have dimension error here

  • @rise4above
    @rise4above 4 роки тому +1

    thank you so much
    can you give me solution of this exo with matlab plaes
    Carry out the training of multilayer networks allowing to estimate the models
    in the following cases:
    1. Function approximation:
    y ̂ = 𝑓 (𝑦) using the dataapproxi.mat data file
    2. Prediction problem:
    y ̂ (t) = 𝑓 (y(t - 1), y(t - 2),…, y (t- 𝑛)) using the data file:
    dataestima.mat
    3. Problem of identification of a dynamic model
    y ̂ (t) = 𝑓 (u(t - 1),…, u(t - 𝑛u), y (t - 1),…, 𝑧y(t - 𝑛y)) using the file of
    data: dataident.mat

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  4 роки тому +1

      Yes. It is possible. However, I need more details to implement and train such a network

    • @rise4above
      @rise4above 4 роки тому

      @@NuruzzamanFaruquis OK Mr
      I have the all data fil.mat

    • @rise4above
      @rise4above 4 роки тому

      @@NuruzzamanFaruquis can you send me your email address please

    • @NuruzzamanFaruquis
      @NuruzzamanFaruquis  4 роки тому

      @@rise4above You can email me at faruquizaman27@gmail.com. Please add 'Neural Network' in the subject, and include your identity (name, designation) at the end of the email.

  • @tkhankhoje39
    @tkhankhoje39 5 років тому

    Delta is coming undefined..