4 April 2017, 12:00 - 14:00
Deep learning has become a hot research topic pursued by academia as well as big companies. Google used it to beat the world champion in the game of Go. Facebook used it to automatically tag millions of uploaded photos with the names of people. Microsoft used it to provide speech recognition. Accenture used it for cyber security. The number of potential applications seems limitless.
In this seminar, Dr Chik of the Anglia Ruskin IT Research Institute (ARITI) will explain the basic concepts of deep learning and how it is different from traditional machine learning methodologies. Three examples will be discussed: image recognition (using convolutional network structure and piecewise-linear activation functions); speech recognition (using recurrent network structure and special identity functions); and behaviour optimization (e.g. how Google trains the computer to play Pacman, using reinforcement learning with an actor-critic loop). Finally, Dr Chik will introduce some software toolkits which will be useful for researchers to use deep learning to solve their problems.