DBTB INT Daniel Golden r

sure daniel goldin company is our Turris title is senior image scientist I'm going to back up for one second I know we're rolling should I be looking at you or the camera you look at me okay okay the most exciting thing is the opportunity to share the work that we were doing an artists which i think is pretty cool but it's nice to have some vindication from the crowd as well I would say this combination really of data and new modeling so I work on convolutional neural networks which are a pretty new concept from the last four years or so and by putting those two together we really can do things that were not possible before so I work on automatic segmentation in radiology images and just before we had enough data and we had the tools like it just wasn't possible to do that well so now that it is it's a lot of wide open opportunities I guess my talk was kind of specific on how we can use it was about one specific aspect of working with kind of messy data and how to fully exploit all the data you have so I hope that someone in the audience might use that technique or a similar one to be able to use all the data in making a model or gang some insight where they wouldn't been able to before i would say practice i would say start into the deep end and you'll be pretty bad at first and you'll get better and better especially working with experienced data scientists would help a lot and then for the work that I do on machine learning practicing on catechol competitions is something that could be very helpful