Speaker 
Marc Deisenroth 

Affiliation 
Imperial College 
Date 
Friday, 21 June 2019 
Time 
13:0014:00 
Location 
Zoom 
Link 
119 Torrington Place G12 
Event series 
DeepMind/ELLIS CSML Seminar Series 
Abstract 
Highimpact areas of machine learning and AI, such as personalized healthcare, autonomous robots, or environmental science share some practical challenges: They are either smalldata problems or a small collection of bigdata problems. Therefore, learning algorithms need to be data/sample efficient, i.e., they need to be able to learn in complex domains, but only from fairly small datasets. Approaches for dataefficient learning include probabilistic modeling and inference, Bayesian deep learning, meta learning, Bayesian optimization, fewshot learning, etc. Highimpact areas of machine learning and AI, such as personalized healthcare, autonomous robots, or environmental science share some practical challenges: They are either smalldata problems or a small collection of bigdata problems. Therefore, learning algorithms need to be data/sample efficient, i.e., they need to be able to learn in complex domains, but only from fairly small datasets. Approaches for dataefficient learning include probabilistic modeling and inference, Bayesian deep learning, meta learning, Bayesian optimization, fewshot learning, etc. In this talk, Marc will give a brief overview of some approaches to tackle the dataefficiency challenge. First, he will discuss a dataefficient reinforcement learning algorithm, which highlights the necessity for probabilistic models in RL. He will then present a metalearning method for generalizing knowledge across tasks. Finally, he will motivate deep Gaussian processes, richer probabilistic models, which are composed of relatively simple building blocks. He will briefly discuss the model, inference and some potential extensions, which can be valuable for modeling complex relationships, while providing some uncertainty estimates, which will be useful in any downstream decisionmaking process. In this talk, Marc will give a brief overview of some approaches to tackle the dataefficiency challenge. First, he will discuss a dataefficient reinforcement learning algorithm, which highlights the necessity for probabilistic models in RL. He will then present a metalearning method for generalizing knowledge across tasks. Finally, he will motivate deep Gaussian processes, richer probabilistic models, which are composed of relatively simple building blocks. He will briefly discuss the model, inference and some potential extensions, which can be valuable for modeling complex relationships, while providing some uncertainty estimates, which will be useful in any downstream decisionmaking process.

Biography 