Jump to Content
Zoubin Ghahramani

Zoubin Ghahramani

Zoubin Ghahramani is a VP of Research at Google, leading Google Brain, as well as Professor of Information Engineering at the University of Cambridge. Before joining Google, he was Chief Scientist and VP for AI at Uber. He served as the founding Cambridge Director of the Alan Turing Institute, the UK’s national institute for data science and AI. He has worked and studied at the University of Pennsylvania, MIT, the University of Toronto, the Gatsby Unit at University College London, and Carnegie Mellon University. His research focuses on probabilistic approaches to machine learning and artificial intelligence, and he has published about 300 research papers on these topics. He was co-founder of Geometric Intelligence (which became Uber AI Labs) and has advised a number of AI and machine learning companies. In 2015, Zoubin was elected a Fellow of the Royal Society for his contributions to machine learning and he is the winner of the 2021 Royal Society Milner Award for outstanding achievements in computer science in Europe.

Research Areas

Authored Publications
Google Publications
Other Publications
Sort By
  • Title
  • Title, desc
  • Year
  • Year, desc
    Plex: Towards Reliability using Pretrained Large Model Extensions
    Du Phan
    Mark Patrick Collier
    Zi Wang
    Zelda Mariet
    Clara Huiyi Hu
    Neil Band
    Tim G. J. Rudner
    Joost van Amersfoort
    Andreas Christian Kirsch
    Rodolphe Jenatton
    Honglin Yuan
    Kelly Buchanan
    Yarin Gal
    ICML 2022 Pre-training Workshop (2022)
    Preview abstract A recent trend in artificial intelligence (AI) is the use of pretrained models for language and vision tasks, which has achieved extraordinary performance but also puzzling failures. Examining tasks that probe the model’s abilities in diverse ways is therefore critical to the field. In this paper, we explore the \emph{reliability} of models, where we define a reliable model as one that not only achieves strong predictive performance but also performs well consistently over many decision-making tasks such as uncertainty (e.g., selective prediction, open set recognition), robust generalization (e.g., accuracy and scoring rules such as log-likelihood on in- and out-of-distribution datasets), and adaptation (e.g., active learning, few-shot learning). We devise 11 types of tasks over 36 datasets in order to evaluate different aspects of reliability on both vision and language domains. To improve reliability, we developed ViT-Plex and T5-Plex, \emph{p}retrained \emph{l}arge-model \emph{ex}tensions (henceforth abbreviated as \emph{plex}) for vision and language modalities. Plex greatly improves the state-of-the-art across tasks, and as a pretrained model Plex unifies the traditional protocol of designing and tuning one model for each reliability task. We demonstrate scaling effects over model sizes and pretraining dataset sizes up to 4 billion examples. We also demonstrate Plex’s capabilities on new tasks including zero-shot open set recognition, few-shot uncertainty, and uncertainty in conversational language understanding. View details
    Deep Neural Networks as Point Estimates for Deep Gaussian Processes
    Vincent Dutordoir
    James Hensman
    Mark van der Wilk
    Carl Henrik Ek
    Nicolas Durrande
    Advances in Neural Information Processing Systems, Curran Associates, Inc. (2021)
    Preview abstract Deep Gaussian processes (DGPs) have struggled for relevance in applications due to the challenges and cost associated with Bayesian inference. In this paper we propose a sparse variational approximation for DGPs for which the approximate posterior mean has the same mathematical structure as a Deep Neural Network (DNN). We make the forward pass through a DGP equivalent to a ReLU DNN by finding an interdomain transformation that represents the GP posterior mean as a sum of ReLU basis functions. This unification enables the initialisation and training of the DGP as a neural network, leveraging the well established practice in the deep learning community, and so greatly aiding the inference task. The experiments demonstrate improved accuracy and faster training compared to current DGP methods. View details
    No Results Found