The solution that we come up with it’s likely to be much more open Then we know the end Is it okay Similarly he introduced this concert of bias here to make this new one richer in evaluating things If I don’t give the bias here then it’s a restricted to using only these two weights and come out of the solution Are you with me Yeah So he was the 1st 1 who introduced the concept of bias.
He was the first Tony introduced the concept of errors basically difference between actuals and predicted on He was also the 1st 1 to introduce off way off reducing these errors by controlling the weights again Linear regression only So in the direction we have S S E this is equal to a white hat minus y rays too poor too Some off creditors Yeah Some means you’ll have a sigma I still the one to end Number of data points you have Then I’ll write This is why I wear with me Okay so we have the summer school dinners Now let’s ignore all this symbols Let’s focus on in these two Why had minus y is the area right now Okay Can I write This is and mix Let’s see Minuses.
Why generate it yeah okay That means in this if this is it here in this enter this editor is influenced X And why come from your data set You can’t do anything with it X And why I come from a data thing It’s party for training Sit There’s nothing you can do about it What you can control is this in on the sea This one these two are want defines your error so your error can be controlled When you control the M and the sea there can be controlled When you control the embassy awaits the raid Socialise em on the sea The cve It goes into your Equus What’s he goes into the equation.
This go with three in Toby’s What goes into the question So w three W W one If I manage with them I can minimize this editor and then you don’t zero Are you with me All of you Any questions So my question is that now with this linear regression the same equation is on the Rosenblatt question Also so does that Whitehead on the Rosenblatt equation will give the same result as that linear regression where why you had will give because that’s the mission Because of this transmission I just wanted to ask one question And since we’re saying that M and C are going to change based on the era but X and fire are different data points.
Which means that for every two inputs of X and y that that whole question will give shift things that were and san was very good observation actually due to get out a question do you understand what you’re saying You’re going to adjust the MNC keeping in view one particular regard at the time for this recording but need different MNC for this recording might need different tendency That’s a question right What’s your end individually Yes they will try toe find the best it light But what they are actually trying to do is find one line that covers all the really response.
They will find error based on one line for one particular data point and then calculate error for all the response that that are available to us Refine which which one little one which will be the one line which will have the least error Okay I think you’ve got me in the right direction But the choice of words could have been different See finally in regulation also what happened Senior immigration Also we need aggression that optimal weight that a stone for your lenient model for some records will continue to give in it for some records will give you good results Your white hat Reese Amos Why But for some records my hair may not be Seamus by even in the best reclines But overall across all the data put together you get the Easter So what drives the M and the sea is overall sum of squared errors not indigent letters So here also really evaluate different weeks and save it and releases that it’ll not indigent record level but overall together Unfortunately this particular gate and other gives some of the other gates those gates the way that data is between inputs and outputs.
We call them Lee Nearly separate data sex Do you know what the meaning off Lena Inseparable date as it is Have you come across this before Linearly separable data sets of those data sets whose classes can be separated our 100% accurately with the linear model on these data sets happened Toby linearly separable data sets I’ll show that to you in a moment because off which it is possible by these algorithms to find one set off m one and m two and see which gives a zero error across all the data points Problem happens when your data sets are not linearly separable in that situation you know to settle for a best fit line where the best freight line minimizes and some moss creditors.
It doesn’t become zero which means ah best freak line will continue to give orders for some of the inputs Good Keep this in mind This is a very important concept which people miss out good Granted your suspicion it’s the best Recline will not give you correct answers for all inputs some input will give a good answer Some important give incorrect answers But the degree of incorrectness across all the data points put together will be the least okay on In this case the given data sets that we have here X annoy their linearly separable data sit and hence there will be one unique combination off MNC which will give you zero for all the edibles Okay so let’s see What is that How does it happen Let’s see that will come to the court and we’ll see that old So one question for the leaning inseparable sets that has been happening is happening to ask before we put the deed in the morning.
It’s our it’s our exercise and we try to do that using a particle The future Engineering and features feature Insulating feature Selection is not the most important tasks that you have to do in your mission landing exercise but in practical in reality you’ll not come across the street Is it’s if you come across this data sets which are very nearly inseparable Who needs a model Why do you need We don’t need Morley right All we need is a room So if you lean in inseparable literacy if something listen this then he’s going to be ready for something more than this is going to be known to fort If you have the factory built with you then why do you need a model you started Sign Reinbold That’ll ever happen But they did have become across really that they never really near misses All right let’s go 10.
These are the gingers necklace and blood brought in This is a remarkable improvement over Mecca Lop its water pits Chino And when he demonstrated this the interested in artificial neuron it went through the roof They’re lot off Expectations on board is going to happen You have videos available from the 19 sixties where some of this data scientists were interview on in their interview just going.
What All the same how the company is going to overtake the human beings and rule the world and all these things happen All the type happened immediately after Rosenblatt came over this new and he demonstrated that it can learn the human hand in all of these Bertram immediately to gentlemen Peppered and Minsky who were working in a mighty land Something at the time these two demonstrated that are so prone This is small personal Rosenblatt is compressible They demonstrated that this perceptron I cannot learn all the hits.