DBTB INT John Whaley r

I am jaan waliye I am I'm sorry I'm jaan waliye founder and CEO of unify ID yeah it was it was a lot of fun because I get to you know work and interact with a lot of data scientists which you don't necessarily get to to every day and hear from a lot of people at different different companies and different organizations you know rather than people working you know working with our tight bubble and yeah i already had just after our talk like we had a lot of interesting conversations with other people yeah i think it's i mean the the number one reason i think is that we now have just a lot more sources of data than we used to in terms of tons of sensors around as of you know just like this explosion of data where whereas earlier is very hard to get access to good quality data and now it's like we're almost drowning in data and we have to decide what what's important verses not so and you know this is really the renaissance time for for data science and machine learning yeah so would hope so what our talk was really about using unique things about each individual and then picking any things like your behavior your environment you know your biometrics things like that and then you're able to sense these just passively using this the sensors and you know in your life like on your phone on the computer you know I OT these things and then just and then just combining this information from these sensors you can actually uniquely identify and not only identify but authenticate people so we talked specifically about 11 aspect which is gait analysis like the way you walk and specifically around your cadence like the number of steps you take and how quickly you walk and it turns out that you know based on your unique physiology there is everybody has a row number about how many steps they take per minute and it's very just with four seconds of data you can uniquely identify that number and thereby uniquely identify the person who is walking and so it's it's a tricky problem there's a lot of you know we're taking the data from the phone there's a lot of accelerometer data there's a lot of noise in there the sampling rate is not uniform you have to take all these things into account but once you do you can actually even extract out of you know a very strong signal about who somebody is just by looking at four seconds of their walking so I mean what like I mean 11 obvious thing is you need to practice I mean like in terms of just just study in practice but and one thing that I think that a lot of day and scientists lose like the is like one factor that they that they miss out on is like really trying to understand like the underlying process like the of what you what's being measured and then if you have this type of insight about what is actually happening on the non drilling process then you can have these very key insights about like you know about the problem whereas if you just took a bunch of random data and you don't know where it came from then like you can't you can't really say very much about it so I think it's important you know as a you know to be a very to be a great data scientist it's not only understand i cannot how to analyze data but also understand like where that data came from like you know is it's from a sensor is it from liking this you know and how is it measured and what are the idiosyncrasies of the way that that happens and if you come to that level of understanding then you can make much deeper insights than just somebody who just had a bunch of random data that if they weren't really sure what it was [Music] you