DBTB INT Carlos Guestrin r

I'm Carlos gastrin I'm the Amazon professor machine learning at the University of Washington and I'm the CEO of data we've reached a phase transition in the world really in the world of machine learning what we're going from just an academic endeavor to the real world being impacted by intelligent applications they use machine learning at their core and in an event like this you see developers and later scientists coming together with the goal of building such intelligent applications the timing for Billy intelligence applications is really right right now so companies have invested time and energy collecting data putting one to one place we're not getting enough value out of it on the other hand the company's herb garden value have differentiated their offerings by doing something exciting their data Abby it amazon in the early days with recommender systems or uber today of how we know did their self in the taxi industry and so the timing is just right to do that and the creativity now can be unleashed to do things we never imagined could be done before with data there is a journey to create intelligent applications that journey starts from being inspired by the possibility of what you can do and how could how can you do that easily and quickly all the way to take it into the production to impact your users or your customers and so there are two things that I talked about today I'm super excited about one is ways to accelerate that journey to help say developers they don't have a motion learning background to be able to do that easily and quickly using the data platform and the second thing is the question of trust how can you know when to trust a machine learning model in production in other words if I have an idea it seems to work how do I know that it's working how do I know it's working for the right reasons and how can I keep doing that as I take the production so that i know that if it's still working still doing what expected to do so if the data platform we're thinking both about accelerating the journey but also providing transparency and clarity for machine learning models so I've been teaching machine learning for a long time and I have lots of ideas of how that can be done but today the way I think about it is start from the use cases start from the applications open them up because those can be really inspired and try to figure out how those were built and were the techniques are support them and from that you can prioritize what the things you can learn and how so with emily fox university of washington we've created a Coursera specialization machine learning that really focuses on taking use cases and opening them up and helping figure out how machine learning works and why you