Stochastic Neural Networks and How to Make One

• What we'll do
The goal with stochastic neural network is to build the next generation of deep learning models which aim to learn more efficiently, and eventually become truly creative.
In this talk, we start with general stochastic models and how they can learn latent factor representations of complex data. Later, we focus on Variational Autoencoders, compare them with deterministic form and try them on MNIST data.

Speaker: 'Najmeh Abiri' is a PhD student at Lund University, Computational Biology and Biological Physics department. Her research is mainly on developing methods to preprocess data and in her research, she has been using deep learning and Bayesian methods.

• What to bring

• Important to know