Monday 30 March 18:30 - 20:00

UCL AI Centre
Darwin Building B40 LT,
London
WC1E 6XA

Tickets Unavailable

UCL x DeepMind Deep Learning Lecture - Unsupervised Representation Learning

Science & Technology

UCL x DeepMind DL Lecture Series - Frontiers in Deep Learning: Unsupervised Representation Learning by Mihaela Rosca & Irina Higgins

UCL Centre for AI is partnering with DeepMind to deliver a Deep Learning Lecture Series.

Frontiers in Deep Learning: Unsupervised Representation Learning by Mihaela Rosca & Irina Higgins

Unsupervised learning is one of the three major branches of machine learning (along with supervised learning and RL). It is also arguably the least explored branch. It had a great start with RBMs used to initialised early deep networks to improve classification accuracy. However, later successes of end-to-end learning meant that unsupervised learning went out of favour. It is now coming back as a potential solution to many outstanding problems with modern deep learning (like data efficiency, generalisation, robustness, fairness etc), with Google Brain recently announcing the Visual Task Adaptation Benchmark to promote representation learning research. We will start by outlining the outstanding challenges in deep learning, before introducing different approaches to unsupervised learning and how the learnt representations have been shown to help on various tasks both in terms of supervised learning and RL. In particular, we will mention disentangling, CPC, VQ-VAE, maybe some of the energy based models, siamese networks etc. We will also discuss the more open-ended question of what makes a good representation.

Bios:

Irina is a research scientist at DeepMind, where she works in the Froniers team. Her work aims to bring together insights from the fields of neuroscience and physics to advance general artificial intelligence through improved representation learning. Before joining DeepMind, Irina was a British Psychological Society Undergraduate Award winner for her achievements as an undergraduate student in Experimental Psychology at Westminster University, followed by a DPhil at the Oxford Centre for Computational Neuroscience and Artificial Intelligence, where she focused on understanding the computational principles underlying speech processing in the auditory brain. During her DPhil, Irina also worked on developing poker AI, applying machine learning in the finance sector, and working on speech recognition at Google Research.

Mihaela Rosca is a Research Engineer at DeepMind, focusing on generative models research and probabilistic modelling, from variational inference to generative adversarial networks and reinforcement learning. Prior to joining DeepMind, she worked for Google on using deep learning to solve natural language processing tasks. She has an MEng in Computing from Imperial College London.

Hide Comments Comments

You must login before you can post a comment.