How to use (NPL) Natural Processing Language?

My focus was more risk in operations and I was leading a team of data scientists after their right so within my previous role I was working on a use case where we were using you know for we were trying to optimize her customer care spend call center to spend.

What we’re trying to do was using natural language processing NLP which is a deepening technique we would figure out and also utilize understanding by understanding exactly you know what kind of activity has a customer done before calling us it’s so basic idea is that people have issues and then that’s the reason they call customer care for an easy resolution right so we try to understand and try to predict that what would be the reason before they called us if they called us.

What could be the reason that disguised calling us and using some of the initial sentences which deciphered end we changed the idea menu so that the first two options would closely match what this guy was the reason this guy’s calling for like and then what we realized was in 60% of the cases where this guy would have otherwise moved to her to speak to an agent the query was resolved in a much shorter time and with no cost because we had the solution ready for that customer so some of these use cases so many companies many businesses have these customer care centers and call centers right.

Now imagine some of these use cases becoming more popular and they become utilizing it they have to write because if the competitor can cut their customer care center by cost by half and if they’re not able to then they will lose the competitive edge so so this is a particular specific use case which I have done using deep learning, but this is it’s a good example on why when every company has to go through this journey and start adopting some of these advanced techniques right and what it means is that the jobs’ creation in machine learning and AI and deep learning will happen first of all it has been happening at a very healthy rate right now.

So in the last five years, we have seen 5x growth in jobs for AI right and ASSAM predicted that in 2018 last year that we have created a hundred thousand jobs in machine learning and AI in India itself right so we have seen that they seem very healthy demand for machine learning in AI but as he’s going on this rate this demand is going to become higher and higher and higher right because of some of these rapid developments and use cases and companies adoption right off some of these techniques okay, so this is a slide which summarizes exactly what kind of trends are emerging in AI right.

I want to you know I talked about the bottom right corner which is for point-like 5x growth in jobs requiring high skills since 2013 in the five years since 2013 we have seen 5x growth right the other thing which I want to point out here is the error rate which is the middle left and the adjacent box has gone down from twenty-eight point five percent in 2010 and in seven years it has gone down to two point five percent right and for Facebook B phase that is now in 2018 it was only 83 basis points right so now with thinking drastic improvement in the algorithms a lot of use cases which as I was describing some of these we cannot go to market with a 30 percent error rate within some of the use cases but you can definitely go to market with any 80 basis points every bit.

So when these use cases are now becoming more and more mainstream right and can now be applied to a variety of business problems so so then that is what is fueling this growth in demand for AI and creating a lot of jobs in AI right the other thing which I want to point out here is that 84 percent of companies have already invested in AI o me that’s a staggering number right 84 percent and I was surprised when I looked at this number because from our popular perception from talking between us from our between appear to grow among appear group right the sense that we would have the others would be of a lower number but 84% is staggering I mean that’s.

We do see that momentum building towards more jobs and more demand for AI and we saw that happening, ut this momentum is building to get more and more of you know these jobs created Nicola right and these are some of the charges which this point which means the same point which I was making that if you look at the trend of AI course enrollment that is just picked up beginning let’s say 2013 2013 2014 you see a dramatic change in the slope right and that’s where these some of these use cases are becoming more and more production-ready.

If you look at ml course enrollment the chart picks up a little earlier it picks up from 2012 right because machine asset business machine learning has been fairly pervasive and I think some of these guys some of you who are in the service sector would also have realized within your work in the kind of projects that you’re you’re getting from your clients these days.

I think that most of these projects are in machine learning there is a marked shift in the kind of projects that are coming from some of these companies right in you would also even otherwise in your workplace you would see that this is a kind of shift which is happening and if you break it down further these are the skills which are come which companies are looking for which is a machine learning deep learning and again we see that there’s growth since only 16 right and it’s just growing at a very amazing place and all of the skills which are in demand machine learning deep learning natural language processing computer vision speech recognition right all of these skills are first of all they are within that realm of machine rolling in AI skills.

These are all the skills which we also teach in our program and you will learn in the program late the other indicator is GitHub stars which basically is you know awarded to people who are contributing meaningfully who get up and get up as a repository where some of the special professionals using machine learning and AI techniques use as a platform to showcase the kind of work that they’re doing right and you see what we see is that TensorFlow which is a library use for more deep learning algorithms in sci-kit-learn is a Python library which is used for machine learning algorithms right.

We see a much more rapid trend or increase intensive flow since 2016 almost starting from the same level as we see that TensorFlow is now growing and obviously the machine learning algorithms also currently and by the way GitHub we in the program be also us get up because the idea is that whatever you do in the program so during the program what each participant gets to do with us is to work with a lot of projects in lab exercises right and all of these are based on simulating some of the business decision or business situations and you know all of the algorithms that you apply you can get a sense of how do you get to apply that in your business.

All of these projects are submitted on GitHub and then each participant builds a repository of projects that he has done on GitHub which again can be used to showcase the kind of experience the kind of learning each participant has and what you observed is you know for this technical community for the community which is working on ml in AI that’s the kind of frame we also see for let’s say for job applications or interviews where people expect to look at your published work and this is a marked change from the previous kind of skill set how the skills were assessed right.

I mean early they used to be a lot of in-person interaction and talking about your experience and all that today the whole shift test awards show me what have you learned to show me what have we worked on what kind of projects have you worked on the right unless discussing on the specifics of that project so having GitHub repository helps in so casing very easily even on the LinkedIn right the kind of projects at your work now and the kind of knowledge that you have in these areas right and the other point is that you know this AI story is not just standing out in the US or in the developed markets it is equally panning out in India like in the last decade the kind of model service model that we have employed in India right.