Introduction and goals

I will start with a broad overview of what neural networks are, and how the back propagation training algorithm works, in theory and in practice.

I will describe some interesting applications, some fascinating phenomena, and some neural network architectures
(convolutional networks for classifying images; transferability of knowledge from one task to another and artistic style transfer; autoencoders; recurrent networks for language modelling; relational networks for reasoning).

I will finish by discussing the role that neural networks should play in data science,
and ask what might come next.

• Speaker: Damon Wischik (CL)
• Tuesday 24 October 2017, 14:00–15:00
• Venue: Centre for Mathematical Sciences, MR4.
• Series: Mathematics and Machine Learning; organiser: Damon Wischik.