Two papers have been accepted to ICML 2016

Two faculty members have their papers accepted to International Conference on Machine Learning that will be held in New York, USA. 

First paper, authored by Anton Rodomanov and Dmitry Kropotov, "A Superlinearly-Convergent Proximal Newton-type Method for the Optimization of Finite Sums" proposes a new stochastic optimization method with fast convergence properties that is especially useful in machine learning problems. 

Another paper called "Meta-Learning with Memory-Augmented Neural Networks" is co-authored by seniour teaching staff member Sergey Bartunov and is a result of his collaboration with Google DeepMind. In this paper a new neural network architecture is developed that is able to quickly learn new concepts from just a few training examples.