15th February 2018, 7:00 pm (event fully booked)
Speaker: Julien Perez
Abstract: During the last 5 years, there has been significant development in the deep learning research community and its application to natural language understanding. Beyond the classical paradigm of supervised learning, the differentiable programming paradigm has started to exploit the notions of memory and attention. Using these innovations, a novel set of tasks are now being investigated like machine reading and dialog systems. This presentation introduces this evolution in the field of deep learning and gives a technical description of gated end to end memory-enhanced neural networks whilst illustrating their usage for machine reading and dialog systems learning.
This event is organised by the Data Science Grenoble group. It will take place at NAVER LABS Europe.