DBTB INT Niyati Parameswaran f

my name is niyati parmeshwar I work as a my name is niyati parmeshwar I work as a data scientist for the IBM Watson group a lot comes from just the name because I a lot comes from just the name because I work as a data scientist and this work as a data scientist and this conference data by the bay is supposed conference data by the bay is supposed to bring in a lot of data scientists to bring in a lot of data scientists whose primary focus and the focus of a whose primary focus and the focus of a lot of talks here revolves around all lot of talks here revolves around all steps with data converting that data steps with data converting that data into information right from into information right from pre-processing munging and then pre-processing munging and then rendering insights out of that data so rendering insights out of that data so that's why it's really exciting to be that's why it's really exciting to be with a familiar group of people it's exciting because it kind of when it's exciting because it kind of when you get really noisy clunky data you you get really noisy clunky data you kind of as the data scientist you kind kind of as the data scientist you kind of want to see what kind of structure of want to see what kind of structure exists in that data so that you can exists in that data so that you can proceed to put in probably a predictive proceed to put in probably a predictive analytics framework or a recommendation analytics framework or a recommendation engine basically find out what that engine basically find out what that structure is so it's kind of a challenge structure is so it's kind of a challenge because every data set that you get is because every data set that you get is obviously not the same and that kind of obviously not the same and that kind of variety allows you to think of variety allows you to think of variations in terms of what you variations in terms of what you potentially could do with that data and potentially could do with that data and how you can leverage it when you're how you can leverage it when you're conceptualizing your algorithmic conceptualizing your algorithmic paradigm or framework I have kind of in my talk I'm talking I have kind of in my talk I'm talking about two projects that frameworks have about two projects that frameworks have built at our Watson and these frameworks built at our Watson and these frameworks while they can be used individually by while they can be used individually by themselves and still have utility themselves and still have utility because one is about face detection in because one is about face detection in belfries generation finger and the other belfries generation finger and the other works for automatic text summarization I works for automatic text summarization I also talked about how we can sync this also talked about how we can sync this to existing Watson services to to existing Watson services to essentially improve the way that the essentially improve the way that the service performs I want attendees of service performs I want attendees of course there are going to be people with course there are going to be people with a clear background in machine learning a clear background in machine learning who would have a sense of auto encoders who would have a sense of auto encoders recursive water encoders and the other recursive water encoders and the other things that I would be speaking of my things that I would be speaking of my talk but I'm hoping that even for talk but I'm hoping that even for someone who was not very familiar with someone who was not very familiar with machine learning I would be able to machine learning I would be able to explain to them from a high level explain to them from a high level perspective as to what this framework perspective as to what this framework does and basically the potential that does and basically the potential that exists when you build frameworks like exists when you build frameworks like these and how they're easy to do it it's these and how they're easy to do it it's not so complicated I'd say the machine learning community I'd say the machine learning community in the data science community as a whole in the data science community as a whole is very very active right now so if you is very very active right now so if you just start from downloading hacker news just start from downloading hacker news one or two out of the five articles that one or two out of the five articles that get published every day would be around get published every day would be around this space just start to just start to this space just start to just start to actively participate and that's one actively participate and that's one space Coursera and Udacity have some space Coursera and Udacity have some really great courses in the space of really great courses in the space of machine learning and natural language machine learning and natural language processing gaggle is something that processing gaggle is something that every mln 2 z's should participate in every mln 2 z's should participate in because they also have tutorials and because they also have tutorials and then you can start if i disobeyed in then you can start if i disobeyed in competition so if you are interested in competition so if you are interested in this case there's a lot of stuff that's this case there's a lot of stuff that's happening around that you can easily happening around that you can easily parenting