Neural Network using Matlab

Please follow me on Facebook: / nzamanfaruqui
Connect with me on LinkedIn: / nuruzzaman-faruqui
Subscribe to my new channel: / @nuruzzamanfaruqui
Contact me on Fiverr: www.fiverr.com...
In this lecture we will learn about single layer neural network. In order to learn deep learning, it is better to start from the beginning. And single layer neural network is the best starting point. This lecture starts with theoretical explanation covering only the essential elements of a neural network. This theoretical section is actually the foundation of leaning deep learning. That is why, it is very important to focus on this section and understand properly. After the theoretical section, a single layer neural network using SGD method has been trained in Matlab. In a nutshell, this lecture covers -
1) Concept of Nodes
2) Concept of Layers
3) Supervised Learning
4) Delta Rule and Generalize Delta Rule
5) SGD, Batch and Mini Batch Method
6) Practical Implementation (Training a Single Layer Neural Network in Matlab)
Most of the time in university class environment students learn the theoretical aspects only. To be honest, it is difficult to cover both theorical and practical aspects in the class. However, this is the best approach of learning anything. And of course, the best of anything comes with a price. I’ve noticed that very few university instructors of my country have the mindset of teaching in proper way - which is the combination of theory and implementation of the theory. Learning without gaining the ability to apply is useless. If you are a victim of these types of poor education system, this lecture will help you to learn neural network properly.
After following this lecture properly, a student will be able to implement single layer neural network in Matlab. The example shown here is done in Matlab. It does not mean that you are bound to use Matlab. Using the concept explained and procedure shown here, you can train single layer neural network using the programming language and IDE you are comfortable with.

[@abdullahullash8471]

You teach people from the very basic which i like most about your contents. Keep it up.

[@eyupburakylmaz7710]

1:08 I thought 'What about Multi-layer Neural Network?' so waiting for that video... When you are able to make and upload that video?

[@NedSar85]

8:40 Generalized single layer equation 12:44 learning rule definition 12:57 Error definition 20:33 Training example in matlab

[@juanbenito95]

Bro next time don't be so obvious, I thought you were really teaching, but I remembered a book I read and realized you were just repeating everything the book said.

[@erajikholiqzoda1120]

Hello sir, I got error a file SGD_method which I have view so it show line d= correct_Output(k); Can you specify for the issues? Thank you in advance. transposed_Input = input(k, :)'; d = correct_Output(k);

[@MsAaradhyaChopra]

Yes same here

[@MsAaradhyaChopra]

I have used cell2mat to rectify the same but now getting error in using - operator

[@erajikholiqzoda1120]

@@MsAaradhyaChopra I cant save Training_Network.mat// I run but still the same. Maybay I import path specific libraries that it can bring it save into current folder?

[@fab10barr0s]

This lecture is amazing! It will help a lot In my graduation project. The only thing that I did not like is that the sound is too low, so I had to download the video to amplify the sound manually. That is the only caveat.

[@NuruzzamanFaruquis]

I am glad to know that it was helpful. And sorry about the difficulty you had to face for the low volume.

[@priyadarsinirajan1248]

I had seen lots of videos about ann but no one explained in such a way as you explained...you are just amazing...looking forward for your more videos on ann..Thank you for making such effort and helps us to understand the concept clearly...

[@NuruzzamanFaruquis]

So nice of you

[@debaleenaroy3685]

Hi. The video is really helpful. As I am a big fan of MATLAB I want to convert my python code to MATLAB. I was just wondering where is the loss function you described? Thanks.

[@AkashSingh-rv8vx]

Same question. Although the SGD must be taking care of the same, the loss function should have been explained for this problem.

[@arunchunaiannadurai2903]

Thank you Professor, Excellent work

[@mortezasaadatmorad3479]

Very useful. Thanks for your step by step teaching. Generally Indian are very powerful in theoretical science.

[@ReBadawi]

Thanks a lot for all the explanation and applied examples. Would I just ask 2 questions: 1) concerning MATLab code I got an error in "Training" script file when I am trying to repeat the Weight function for 10000 epoch as I got [ Error in Training (line 13) .. Weight = SGD_method(Weight, input, correct_Output);] and I don't know from where the error could be . 2) Why we add a semicolon at the last line of initialising 'input' matrix before we close the brackets which is in line 4 in code explained in "Training" file script as well. Great thanks.

[@catherinemukunga6458]

now i can write by methodology in confidence....Thank you sir

[@NuruzzamanFaruquis]

Yes. A strong basic gives confidence. I am glad that my video helped. Please subscribe and encourage your friends to subscribe my channel.

[@skndshaum]

PLS help! .. it´s not working :( The error message: %%%%%%%% Error using * Incorrect dimensions for matrix multiplication. Check that the number of columns in the first matrix matches the number of rows in the second matrix. To perform elementwise multiplication, use '.*'. Error in testing (line 11) weighted_Sum = Weight*transposed_Input; %%%%%%%% And if I rewrite the code: weighted_Sum = Weight.*transposed_Input; .. then I get the next Error message: %%%%%%%% Error using / Matrix dimensions must agree. Error in Sigmoid (line 2) y = 1/(1+exp(-x)); Error in testing (line 12) output = Sigmoid(weighted_Sum) %%%%%%%%

[@jacktimber6153]

Thank you sir. Your lecture is very clear to understand. Please make more neural network tutorial.

[@faizakanwal4206]

Hi. I need a guidance for my research "Encryption and decryption of a 6-bit plaintext using multilayer neural network". I'm getting errors in my code and can't really figure it out. Anyone can help with that?

[@merondesta3895]

thankyou for the video that was helpful. do i need neural network tool box to run this? weight = SGD_method(weight, Input, correct_output); it kept telling me that there is an eeror here

[@dawitmusse3548]

Thank you. But I have one question. How do we incorporate biases to this model? How do we initalize and iterate for biases?

[@AliHussainSyedAP]

Hi sir what ever the code you wrote for the sigmoid function wont run by matlab bcz sigmoid is the function were we have to define the ranging from were to were our activation function should confined, please can you clear my doubt sir, Thank you

[@theengineered1949]

Your pauses are too long my man

[@NuruzzamanFaruquis]

Thank you for the feedback. I will try to better the quality of my tutorial.

[@DeepakSai10]

U have got the trained outputs. But can u explain where exactly the weights are changed and at which node using MATLAB?

[@NuruzzamanFaruquis]

I will try to cover it in some upcoming lectures.

[@Krantiveer-y7u]

thank you so much sir ....sir plz provide ppt of those lecture on neural network.

[@Shashikant-sz2sw]

Sir could you please make a lecture on recurrent neural network (RNN), SVM and Echo state network (ESN)

[@NuruzzamanFaruquis]

Thank your these suggestions. I will try to cover all of these

[@salmanx9923]

Error in sigmoid (line 2) y = 1/(1+exp(-x)); sir please guide me

[@alihassanawan7445]

kindly describe the alpha value which you give in the code is 1.....

[@achyuttrivedi6296]

Hello sir, I am facing an error as K is undefined. I follow your exact procedure, sir. Please help!

[@NuruzzamanFaruquis]

I think you used 'capital K' and 'small k' in two different places as same variable.

[@ahsanshaikh5236]

Thank you so much for this detailed explanation. Keep it up sir .. !!!

[@NuruzzamanFaruquis]

Most welcome!

[@tsehayenegash8394]

How can I predict temperature data by using neural network in matlab progam

[@darsankumarpj6413]

I'm getting all output as 0.5. No errors shown

[@MrSaviotp]

Great lecture, thanks! Please make more videos of neural networks in Matlab.

[@dimasrayindra6468]

why my output not appear? i try 3 times exactly the same at the video but still nothing

[@peddirajuradha]

Very nice explanation includes both concept programming .Thanks for sharing .

[@NuruzzamanFaruquis]

You are most welcome

[@rockagovkal9065]

Can anyone share me this code mine is not working showing that it has error while calculating sum

[@PandiyanM22PHD064]

how to do like this for bearing fault classification

[@mudasirjamil9942]

I am getting 0.9+ values in the output. Even though correct_output is set to [0 0 1 1]. I have verified the values multiple times... Can you suggest what could I possibly be doing wrong ???? Thanks

[@add-wisehaalathakikatkeanu5342]

How to save the NN after its creation; hw to use it later on for ex for cross-validation

[@muhammadashraf6645]

Can I have your email,sir?? Plz

[@muhammadbudid7693]

Sir, can you explain again why the formula of weight equals to 2*rad(1,3)-1, I still dont know yet where "2" comes from. Thank you

[@carda1917]

To the SGD-Method: If I wanted to train a neural network that approximate a function and if I have 50 data points I need also 50 weights?!

[@NuruzzamanFaruquis]

I am sorry I didn't get your question properly.

[@carda1917]

@@NuruzzamanFaruquis It's fine. I got it. The SGD-Method you implemented and explained - Isn't it just the GD? I didn't get where you just take a random part to calculate the gradient

[@rafaaracabral]

`Can u tell us the bibliography?

[@saifullahkhalid5162]

I think there is some issue in training.m. I had to change Weight= 2*randi(1,[1,3])-1; to run it correctly.

[@NuruzzamanFaruquis]

I will recheck. Thank you for the comment.

[@muhdfaris5859]

can matlab build an deepfake detection???

[@azmanman3454]

Thanks for such wonderful video, the explaination is superb and easy to understand. I have 1 question, during initializing the weight , the formula is weight = 2*rand(1,3) - 1 , why needs to times 2 and minus 1 ?

[@NuruzzamanFaruquis]

To make sure getting non-zero and non-negative values. Thank you for watching my videos.

[@azmanman3454]

@@NuruzzamanFaruquis Thanks sir

[@Ab.zebi12]

i needthe next part of this video multilayer neural network

[@suleimanbashiradamu]

Very rich presentation and informative. Thank you very much.

[@NuruzzamanFaruquis]

So nice of you

[@debashisbiswal6798]

Well explained every concepts and that programs through matlab. Yes i subscribed.

[@dodgeridersrt5650]

If i want to apply this but my input is images ... how can i feed it to the neural network ? Please help :)

[@ghaliahmed]

good job Excellent introduction to neural network basics, keep up the great work! very good friend! keep going .....

[@SandeepGaikwad]

This is one of the best video of ANN. Thank you so much, sir, for clearing my doubts

[@NuruzzamanFaruquis]

You are most welcome

[@knowledgeclub4441]

Hi I need your assistance please

[@abdousli2670]

Thanks you for useful méthode if explanationشكرا

[@khurramaziz2430]

Please make more Lectures on Neural Networks using MATLAB ASAP

[@rothchanmithona8642]

Appreciated your hard work. Anyway I had a question "How to know the correct output of the input" since using SGD method mean we already knew the correct output. If I want to use ANN to find the correct output of any input without knowing the correct output which method should I use??? Thanks in advance.

[@NuruzzamanFaruquis]

Thank you for the appreciation. If the output is unknown, then you have to use unsupervised learning.

[@rothchanmithona8642]

Anyway do you have an explanation video about unsupervised learning in your Channel??

[@NuruzzamanFaruquis]

I am sorry to say I do not have any video on unsupervised learning.

[@rothchanmithona8642]

Thank you anyway 😊😊

[@aaronmailhot9370]

Thank you for the GREAT introduction video Sir, your content and pacing are excellent! I did have one point of confusion I was hoping you could clarify in your 2-layer example at approx. 8:30 in the above video: You describe the second row of the weight matrix [3 1; 2 4] as being 'the weights of the second layer' (which your annotation also shows), but these two numbers do not make sense from my perspective. I believe that rather the weights of the second row ([2 4]) are the weights between the inputs and the bottom node in the first hidden layer. I.e. I think your further explanation at approx. 8:50 is correct on this material, but it is just incongruous with your explanation at approx. 8:30. Regards and thanks again!

[@silvalorraine]

That's true! I guess, he means that the weights of the second node i.e [2 4] in the first layer are taken as the second row.

[@geomatictutorials5526]

Am impress by the way you explain. keep it up Sir

[@NuruzzamanFaruquis]

Thanks a lot

[@ShafiqIslam]

very good... Did a fabulous job. May Allah bless U !

[@nanazeethiopia2892]

Very Nice! Thanks a lot. Your presentation is amazing !! Keep going please.

[@michael498176]

Great tutorial! Is the algorithm able to accurately predict the output using a different input from the input in the training set?

[@NuruzzamanFaruquis]

As long as it is possible for a single layer neural network, yes, you can use other values. However, single layer neural networks have limitations.

[@zahraakhalid4230]

سلام عليكم بس هاي مال xor لو لا؟

[@youghi]

can we solve the iris flowers problem with this way you described here ?

[@eliech6411]

Can you make an RBF network example please?

[@ayadsobhan6195]

Sir pleass i need heelp me pleass

[@bismasaqib7143]

I want to make a report on machine learning using the matlab software I m ready to pay

[@NuruzzamanFaruquis]

You can email me at [e-mail]

[@souravb613]

Great to know that you are a bengali . I m too. Liked your video very much.

[@ummeynayema980]

Excellent idea share 🥰 Thank you

[@NuruzzamanFaruquis]

Thank you.

[@akramdawood2477]

thank you ,your explanation is great

[@NuruzzamanFaruquis]

You are welcome!

[@FF-fu9nr]

I didn't understand why you wrote: Weight = 2*rand(1,3)-1; in 25:22...

[@ajeetkumar-ky5ek]

I think you should teach English xcent to weak students like me. Some super star become all black and make astonished to other. Mind blowing explaination.

[@NuruzzamanFaruquis]

Thank you for your inspiring comment. I wish I could manage some time to prepare lessons on English accent.

[@imranyar5951]

Wonderful lecture ...sir I did it but the result produce is not matching with desire result. My Nnetwork generate 1 in every case what could be the reason ?

[@michaelmwasela5249]

Bro, your content is amazing, thumbs up

[@nalinikarchi9350]

trained_network .mat is saved as trained_network.asv

[@NuruzzamanFaruquis]

The '.asv' is a autosaved file. You are mixing up the '.mat' and '.asv' files.

[@ИсмоилОдинаев-й2я]

Thank so much!!!! I found what i wanted to found.

[@NuruzzamanFaruquis]

Glad I could help!

[@አራጌደጉ123ቻናል]

10Q

[@ezzianehassane1127]

Thanks Dr for explanation ... Please, I want a code of neural networks to check the diagnostic faults . We have an electrical transformer and it has many faults And every corruption has its own behavior Suppose we choose four types of faults and each type we try 50 times and each time we take the data for that attempt (we have 200 datas). We enter these data (for example, we enter 160 data and leave 40 for testing) Then we train the neural network on these faults (data entered), but we enter the data according to the fault pattern, for example: the first damage: 40 data and the second damage: 40 data, and the third damage: 40 data and the fourth damage: 40 data. After the training process We are experimenting with new data from 40 which the rest we have Then we ask the neural network to identify it and classify it for any type of malfunction (is it a type of first, second, third, or fourth damage) Please answer .

[@komailmehrgan9359]

TNX 😍

[@ananthakrishnancg6285]

Super class....understood well..thankyou sir

[@bishnukumar5790]

I love this ..

[@nelsontraquinho981]

Good explanation, it will be a pleasure to meet sir. God bless you

[@duvandeswantara9111]

why the implementation in matlab does not use bias?

[@NuruzzamanFaruquis]

I didn't use it to ensure simplicity.

[@NoureddineBelhani]

thank you very mach it is verry helpful

[@TylerDurden-zv2lt]

You are my new hero. God bless you bro

[@NuruzzamanFaruquis]

Thank you.

[@cayanaraycaudhuri]

Thanks. Bujhte subidha holo. Jodi time series nie kono lectures koren bhalo hoe.

[@chih-tengshen9587]

Thank you very detail. :)

[@chandrakant9122]

how can I download and install neural designer software?

[@Futureworld-AD]

،I have problem with prediction Could you help me This problem While training, R2=1 turns out. Testing is very low. How can to prevent that

[@nag9502]

Thank you very much sir, your video lecture are very useful, if you have video lecture on multiple object detection using deeplearning, please upload.

[@sanaekhanboubi6436]

great thanks for you for your effort to transmit all this information with us and with simple methode i sooo like and it's help me to understand the basics of NNA , i wanna just ask you these qst: why you are repeating the inputs matrix in testig script. and if we add the error function and delta instruction dweight we'll found by end a convergence closer to the correct outputs.

[@NuruzzamanFaruquis]

Glad it was helpful! However, I couldn't understand your question. Would you please be for specific?

[@ChandanaGouriTekkaliPHD]

Thank you for such a nice explanation.. And I have a doubt regarding deep q learning.. I am doing research work and working with Deep Q Learning.. Can I apply dataset ( with 300000 records) to DQL? Sir , I need your help to clarify my doubts. How Can I Contact you in this regard.

[@hadishafiee235]

Excellent explanation for single layer of Neural Network

[@josemanuelvasquezrequejo9757]

excellent, thanks for share

[@NuruzzamanFaruquis]

Thanks for watching!

[@amitdravid4383]

Very, nice explanation

[@NuruzzamanFaruquis]

Glad it was helpful!

[@raminmdn]

Many many many videos and tutorials claim to teach neural networks from the very basics, but they fail to get even close to what you did in this video. Please continue the good job.

[@NuruzzamanFaruquis]

I am truly inspired. Thank you for such inspiring comment.

[@KZ.Candle]

How cool is that ,you are a mazing ,thanks sir (dhanyavaad bhaee)

[@saritpravasahoo4589]

Sir, when we used nntool for prediction

[@NuruzzamanFaruquis]

I didn't get your question.

[@austinetoochi176]

This is brilliant, to say the least.

[@youghi]

thank you very much for your lecture , it was awesome .

[@jayalaxmics8239]

sir how to get external input

[@NuruzzamanFaruquis]

I am little confused. Would you please explain what do you mean by 'external input'?

[@danielzhang3070]

It is not SGD, it is vanilla gradient descent.

[@zakars2265]

Waw man, think you so much

[@peddirajuradha]

can u explain how this eqn is taken weight = 2*rand(1,3)-1

[@NuruzzamanFaruquis]

It is explained in the video.

[@AmjadKhan-do8jf]

Sir, I have few questions If u may guide me I ll be obliged: 1.How many iterations will it require reach the actual value to the desired value 2.What will be the value of weights at the termination 3. How I ll plot the error graph please help

[@Rahulpal-mx7hz]

baki sb achha hai pr fake accent use na krte to or achha hota.

[@NuruzzamanFaruquis]

मेरे भाई, यह मेरा वास्तविक खाता है।