# How to train network security machine learning algorithms?

What is my job now is my job is to train network security that when I give when I say that wh White Okay in English spelling It should also give me the same numbers You know what In the form of numbers And then we’ll map it up in the phone off digits starting in the volunteer leaders All right Now let me show you what exactly are these calculations and what we have done before that how many in boots do I need.

If if you don’t But I say the total number of independent columns is equal to or in the course in this, in this case, can I say I just have one independent collar which is an agreed So I will need only one in How many boots do I need It is Okay First of all what is it Isn’t that a great start on the classifier It is a classical Um look at the numbers Okay Whatever You’re right Because we are passing ABC But in this case, we do not have softmax in excel If I So what I’ve done is I’ve made a regressive and then, later on, will we will try to classify but perfect hands So it’s a classified with one out.

This is one input one All right And then you may ask me Ok what are these other things These are my Children layers you would see down the line water these and how I designed this All right now let’s try to pull one number Say I pulled out W what is we Quando 0.847 Do that is what my actual values this is my actual Why All right now what I’m doing is I have to choose some random rates Do people remember we talked about rates randomly chosen data Now let me randomly through somebody So I will make its a pointed for five for do 45 say 80 67 some random number whatever we like All right.

I have chosen some number as my rates Now what I read we do my weights as my rates will lie on each of this line So this particular 8.7 that you are seeing is a whale given to the slain So when my import 0.847 toe goes to this particular neuron it gets multiplied with this week Are we clear under that Everybody What are these rates This is my biasing factor And this arm actually which Okay base to let me know Are you okay with this Any any shoes Neither Protocol with the contract train with Justin just wanting I don’t understand how this random number comes because we live in We say that our foot is dependent on the input as well as the bait which is coming up.

The way the number which was chosen for the way it is something which will influence out so exactly perfect So what does that job then Toe backpropagate and it just the way It’s such a way that my actuals and my prediction should match focus So basically that’s part of learning in this case So Okay exactly The whole game here is toe justice Wait Such a very bad the output did we get here Matches my All right you see now output and put matching Not at all Right Look at the lost function that I put our embassy That is around 5000 plus or something Agree So whatever We’s so sorry What is the problem that we’re staking here Uh the actual and the limousine What is that way Okay.

The first thing I did was e my deployed this with my current weight and activation function I’ve uses sigmoid under that Now you have You may ask me from where I got sigmoid Very simple We order ah logic here or a macro over here to simulate signaling Now what is sigmoid on A very simple scale Sigma Reuters defined by one upon one plus e power off minus X What is my ex X is one of my inputs given to the signal That’s so what I’ve done is I have my own sick mortal function.

What I’m doing if you observe the formula bar E five in do this one Okay Got it, Yep Minus, my uh this one already say bye ass factor And I see my deduct That means I Whatever manipulation I’m going down the line I am putting a sigmoid onto that And when I put a sick more than two That the value that I get a 0.1 to which I have perfectly spoiled it up Because my original values 84 and the value which I do after multiplication is little beard Okay then So this is my end one hone Same way I take this up multiplied with next wait Protesting modern to that and get this Okay Code one is uh anyone edged So this one.

If you remember If you see here now we’re getting inputs from to off them We’re getting an input out off into a trunk And also the ability Same thing Whatever goes out gets me deployed with the new weighed three We’ll put a Sigma Toronto that as activation in Trudeau All right so our data is propagating from here to here So when my this particular data propagates from here to the output it becomes minus 42 And if you ask me why it is all because off junk rates that we have put agreed everybody agrees to that the weights are causing this issue right So what What we should do now If it is responsible for making this issue We’re supposed to go back and change the babes.

Now how do we do this In neural networks we call it back Propagation will continue its formula now But for now in Excel File what do we have is me inside Data Were something going solver I don’t know A few people have used it My intention here is to reduce my RM Easy Now Everybody knows what is our embassy right Rood means square error It’s a lost function What you mean by lost function Difference between my output predicted output and actual That is what we call is lost Everybody is okay with this adamancy All right Now what is my intention here is to focus on my intimacy here And I intend to reduce my intimacy Agreed Yes I know Yes Yes We’ll go to the formulas now After this real I’ll show in Delaware Do we How to do that Next What is my intention here What I need to change the baby that my armies should reduce Can I say all of these junk wits I have given those raids over here.

Now let me console We’ll take some time But you observe the rates changed and the rigs when the rates are changing Now can you look at my our embassy from 5000 are It has come down to almost zero You agree to a point doing pretend part of minus six is almost zero Yep And look at my output now and look at my important can I say they’re almost in sync Yes or no So can I say I have designed a network Yeah Such a way that tomorrow if I give this particular value it is going to identify that it is the blue.

At least if I give this minute is gonna identify it Are you able to visualize that enough Let destined Let’s try to test its A For example I have given a value s Here s is what What is the value of s 0.6931 Okay so when I give the value automatically all the calculation certain and output that I get is 0.69040 How I got this using these weights and the other that I currently get zero points directly negligible So this is your forward propagation and what we did here Using solar was their backward propagation One forward love one backward three presents one e book in a new in it So nearly it worked continuously goes front and back front and back front and back Till I find my lost function to be minimal Still my rates are adjusted to the very best bangs This is the high level overview often unique So on the solver Ah whatever He did it So it is very very easy and quick to understand Um the soldier was just tryingto um just the rates and keep people trading Is that probably driven Whatever Hatred Any box to get the right.