In this talk I hope to give a clear overview of the opportunites for applying Thompson Sampling in machine learning. I will share some technical examples in recent developments (for example Bayesian Neural Networks using Edward) but more importantly I hope to trigger the audience to start thinking strategically about how we want our machine learning models to learn from new data.
Thompson Sampling was formulated in 1933, it's only in recent years that the machine learning community is starting to fully appreciate it's simplicity and power. This talk will be combination of some technical examples on how you can use developments in modern machine learning to start using Thompson Sampling for your machine learning problems (for example Bayesian Neural Networks using Edward). However, more importantly it should give a very accessible story triggering the audience to start thinking not only how to estimate machine learning models, but also to think strategically on how we want our machine learning to learn from new data. I hope to be able to discuss more general machine learning models like image processing, but I would like also to especially focus on a more theoretical example of the application to medical research.