General Training Session: Generative Adversarial Networks: Models that Create

Abstract: Generative Adversarial Networks (aka GANs) give you capabilities that many people didn't know were possible. Conventional deep learning models can make predictions about objects like images, videos or molecules; But GANs create realistic new objects with desired properties. The power of GANs has caught society off guard as some experts use the models to create synthetic images and videos that fool human observers. In this workshop, you will use cutting edge TensorFlow functionality to create your own GANs, and you will gain enough hands-on experience to later apply GANs on your own applications.

Bio: Dan Becker switched his focus from statistics to machine learning in 2012 after finishing in 2nd place (out of 1353 teams) in the $500,000 Heritage Health Prize on Kaggle.

Since then, Dan has done data science consulting for 6 companies in the Fortune 100, and he has contributed to the TensorFlow and Keras libraries for deep learning. Dan runs Kaggle Learn, which aims to provide the single fastest path to learning advanced data science.

Open Data Science Conference