Based on that, it finally gives an overall probability of that sentence in terms of which currently belongs to Prayer is that is already known to our prayer and they start the exercise when they start the exercises Let us say I do a classification I start with whether positive negative difference on many features and if I have 14 days whether observed nineties are positive that the fight is a negative When I started to exercise my prayer problem is at 5 49 Before that is prayer Then I do a post idiot The posterior probability is something which is the contribution of based here Um and what is important is C given the X is determined from the X given see So you x given.
See I’m using it but my impressed ISI has given X so we’ll talk That inward thing is the contribution of based there now many situations We have this I want to introduce a new product I have some success rates in the past Then I want to find out whether I should introduce this I have some past sexist rates which are private properties But if I introduce I may lose some money and all that At the same time I’m a miss an opportunity I go to a market of such form to predict the success rate And Marc it was such fun When I go I have some pastor cards off their performance How many times the agency has predicted successive product when in reality it is a sexist so probably re agency.
Is that data I have But what they need is a probability that the product will succeed given the agency predicts success Alta This is the singular contribution of based you know and that is what they’re saying See given the X his ex given C X given to see is the probability I know but I won’t see given X So the usefulness up given X will outweigh X given see but X given we told the help of X Given See I cannot find or seek Immunex That is the discovery of based here the famous mathematician of the 18th century and even today we are doing based regression So many things that based their um application and mathematical properties of their even integration were using it Beijing probably analysis It is a powerful revision of probability in the late of information available to you based on the information they re ways the probability rates So if I drilling operation I get 40% chance.
I will strike I’ll before the start of the exercise by using the Beijing analysis of additional information Give my dealing probability Producing oil becomes 48 or 50 That means it is That is a power it gives you Then I will say dealing is what So based on the view that so a geologist comes and tells I will predict whether this oil is not on the charges $1 million Our somebody comes and predictably after quick So I have a plant nuclear plant in the next five years So you have two quick strikes This plant will be terminated and you have to produce a map So I canceled ideologists What is his success rate And is it worth to paying him $1 million for this prediction.
What happens So the base Tatum comes there So that is it sewn A base is one of the places it approximates the based here um and makes it simpler The woods it gets classified in terms of their issue This is the one where the important question when you say accuracy is 74.69 I give you at a classic situation which might challenge you Supposing you are vocabulary but you’re built fa feature that’s not in-floor a meaning that would have been produced by another word, Okay but I meant the same thing that set us It’s this anonymous work on you were limited feature list may put it in the other category, As a result, it showing a wrong accuracy So how do you overcome unless you have the entire dictionary TThere rearrest comparisons and all that This is a real challenge and you cannot do otherwise The I think other ways you should restrict everybody that you should use only these wars.
As I didn’t mention the Vadis algorithm keeps getting more powerful is with more known The algorithm is powerful as the limited vocabulary which you’re built-in the feature set so the features it’s keeps getting enhanced were very varied gets keeping enhanced every time you’re ending more leader to our corpus No no supposing I put a sentence Suppose they say I anticipate the situation That is the vocabulary that is they expect the situation it doesn’t get in is and desperate and puts it some bad Which is wrong But you should That means you’re feature set vocabulary Must be a dictionary Are exhausted treaties some exactly What I give you Another example Is the situation harmless Isn’t harmless Somebody’s is a city Not course So in articles is not there.
It was wrong You’re at your regular civil war down Uh 222 Approaches to this one is we are using the considerable amortization Yeah wwhereinwhichever text is finally being put him to predict the breakthrough granulated to its base words kind of constructions Second is VR only extracting the three types of words which can come in the world addictive and the world’s suburb objective and Advil These are the only three things which were picking on from a sentence If there’s any kind of other words now anticipated expected rest for the indisputable is the case I’m just saying which are the controversy anticipate Expect will be that he apply limit ization to some of these words they get to be baseball.
But in this case also if there’s a word which is not completely handed across there is a possibility of that If my normal I’m saying is your predictive accuracy which gives here could have been low if the character words I’m not are not taken that one of the reasons why the corpus which was building is And since this is getting used in this particular case in the medical lab company the kind of replace which can be anticipated for this particular chat poured application That’s the kind of corpus which what they what where the plan is also is in the future as and when the application gets more robust and more useful to start using it Whatever is the critics’ entries which coming those keep getting at it back as corpus the sister.
Training every enhancing the ability to predict better things assimilate comes in What they do have is a filtering process also where they’re seeing that kind of number off the sentence and the number of words Because we’re only adding these things That kind of thing has been getting a checkup Okay before that application was your question These to your questions gets address because they’re separate More deals which you have made You handle that kind of scenarios So there is a case of negation negative.
I’m not feeling bad if you use it in a D Fort algorithm Based on probabilities is we look to say that it’s a negative sentence because of the presence of the world bad and not so What we have done is we have created a separate more dune for handling negation of negative type of sentences which takes care and actually tells you now that this is actually not negatives, In other words, a positive similarly over 100 negation of positive models wherein somebody might say I’m not feeling good but because of the presence of these words on whether the way by Singlish takes place it will convert this into a positive sentence.