Dr Chris Jewell, Senior Lecturer in Epidemiology and Lancaster's CIR (Computer Intensive Research) Lead on the N8 will give us an introduction to computing with TensorFlow.

TensorFlow is a library for high-performance computation, that provides a simple high-level user interface for expressing calculations whilst providing the ability to transparently run on a wide range of accelerated hardware. In short, if used correctly then TensorFlow can accelerate both your development and computation! In this workshop, I will describe the TensorFlow framework, and introduce the functional programming paradigm it encourages through some simple examples in Python. Moreover, I will discuss the auto-differentiation features of the library (useful for optimisation procedures), and the performance enhancements it offers through graph-mode and "XLA" just-in-time compilation. Please note that this talk focuses on the basics of TensorFlow as a high-performance computing library. I won't discuss the neural-net machine learning capabilities or probabilistic programming framework -- they will come in future seminars!