Computational Perception Group

home | members | publications | directions | research | presentations | teaching | vacancies | personal


The Computational Perception Group is run by Richard Turner. Its research spans the following areas:

  1. machine learning which provides a theoretical framework for learning and making inferences from data in order to make decisions.
  2. computer perception and cognition which builds automatic systems for processing and understanding images, sounds and videos
  3. statistical signal processing which uses techniques from statistics and machine learning to develop new signal processing methods
  4. machine learning for climate science which employs advanced machine learning methods to improve predictions for climate modelling and forecasting to facilitate decision making

Particular current focusses of the group are: deep probabilistic learning, human-like learning, continual learning, k-shot learning, active learning, transfer learning, reinforcement learning, concept learning, probabilistic models, Bayesian statistics, Bayesian neural networks, Gaussian processes, spatio-temporal modelling, approximate inference, scalable and distributed inference, Monte Carlo methods, variational methods, expectation propagation, and Bayesian optimisation. We are also interested in the connection between machine learning and computation in the brain.



opening image


Selected News

28 / 09 / 21 3 papers accepted to NeurIPS2021: Memory Efficient Meta-Learning with Large Images, How Tight Can PAC-Bayes be in the Small Data Regime?, and Collapsed Variational Bounds for Bayesian Neural Networks
25 / 09 / 21 A new journal paper accepted for publication: Convolutional conditional neural processes for local climate downscaling
01 / 10 / 20 Richard Turner Promoted to Full Professor
26 / 09 / 20 5 papers accepted to NeurIPS2020: Continual Deep Learning by Functional Regularisation of Memorable Past (oral), Meta-Learning Stationary Stochastic Process Prediction with Convolutional Neural Processes, On the Expressiveness of Approximate Inference in Bayesian Neural Networks, Efficient Low Rank Gaussian Variational Inference for Neural Networks and VAEM: a Deep Generative Model for Heterogeneous Mixed Type Data.
01 / 09 / 20 Siddharth Swaroop wins a 2020 Microsoft Research EMEA PhD Award
01 / 06 / 20 2 papers accepted to ICML2020: Scalable Exact Inference in Multi-Output Gaussian Processes, and TaskNorm: Rethinking Batch Normalization for Meta-Learning
06 / 01 / 20 1 paper accepted to AISTATS2020: Independent Subspace Analysis for Unsupervised Learning of Disentangled Representations
20 / 12 / 19 3 papers accepted to ICLR2020: Convolutional Conditional Neural Processes (oral), Continual Learning with Adaptive Weights (CLAW) and Conservative Uncertainty Estimation By Fitting Prior Networks
18 / 10 / 19 Richard Turner awarded an EPSRC Prosperity Partnership
03 / 09 / 19 3 papers accepted to NeurIPS2019: Practical Deep Learning with Bayesian Principles, Icebreaker: Element-wise Active Information Acquisition with Bayesian Deep Latent Gaussian Model and Fast and Flexible Multi-Task Classification Using Conditional Neural Adaptive Processes (spotlight oral)
02 / 09 / 19 A new draft paper on Pathologies of Factorised Gaussian and MC Dropout Posteriors in Bayesian Neural Networks
01 / 09 / 19 Dr. Turner becomes Director of the Machine Learning and Machine Intelligence MPhil.
01 / 03 / 19 The new UKRI Centre for Doctoral Training in AI for the study of Environmental Risks (AI4ER) is announced. Dr. Turner is a co-Director.
23 / 11 / 16 Gaussian Processes: From the Basics to the State-of-the-Art, Imperial's ML tutorial series [ slides | video ]
19 / 05 / 15 Richard Turner awarded a Cambridge Students' Union Teaching Award for Lecturing
more...
 
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback.    Privacy policy