News

• OxCSML at NeurIPS 2021

  The group is participating in NeurIPS 2021. Please feel free to stop by any of our poster sessions or presentations! We have 25 papers accepted to the main program of the conference:

  • Online Variational Filtering and Parameter Learning by Andrew Campbell, Yuyang Shi, Tom Rainforth, Arnaud Doucet
    • Poster session: Wed 8 Dec 12:30 a.m. PST — 2 a.m. PST
    • Oral presentation in Generative Modeling: Tue 7 Dec midnight PST — 1 a.m. PST
  • Fractal Structure and Generalization Properties of Stochastic Optimization Algorithms by Alexander Camuto, George Deligiannidis, Murat A Erdogdu, Mert Gurbuzbalaban, Umut Simsekli, Lingjiong Zhu
    • Spotlight presentation: Tue 7 Dec 8:30 a.m. PST — 10 a.m. PST
  • Deconditional Downscaling with Gaussian processes by Siu Lun Chau, Shahine Bouabid, Dino Sejdinovic
    • Poster session: Tue 7 Dec 8:30 a.m. PST — 10 a.m. PST
  • BayesIMP: Uncertainty Quantification for Causal Data Fusion by Siu Lun Chau, Jean-François Ton, Javier González, Yee Whye Teh, Dino Sejdinovic
    • Poster session: Tue 7 Dec 8:30 a.m. PST — 10 a.m. PST
  • Provably Strict Generalisation Benefit for Invariance in Kernel Methods by Bryn Elesedy
    • Poster session: Tue 7 Dec 8:30 a.m. PST — 10 a.m. PST
  • Self-Attention Between Datapoints: Going Beyond Individual Input-Output Pairs in Deep Learning by Jannik Kossen, Neil Band, Clare Lyle, Aidan N. Gomez, Tom Rainforth, Yarin Gal
    • Poster session: Tue 7 Dec 8:30 a.m. PST — 10 a.m. PST
  • Neural Ensemble Search for Uncertainty Estimation and Dataset Shift by Sheheryar Zaidi, Arber Zela, Thomas Elsken, Chris Holmes, Frank Hutter, Yee Whye Teh
    • Poster session: Tue 7 Dec 8:30 a.m. PST — 10 a.m. PST
  • Vector-valued Gaussian Processes on Riemannian Manifolds via Gauge Independent Projected Kernels by Michael Hutchinson, Alexander Terenin, Viacheslav Borovitskiy, So Takao, Yee Whye Teh, Marc Deisenroth
    • Poster session: Tue 7 Dec 8:30 a.m. PST — 10 a.m. PST
  • On Pathologies in KL-Regularized Reinforcement Learning from Expert Demonstrations by Tim G. J. Rudner, Cong Lu, Michael A. Osborne, Yarin Gal, Yee Whye Teh
    • Poster session: Tue 7 Dec 4:30 p.m. PST — 6 p.m. PST
  • Implicit Deep Adaptive Design: Policy-Based Experimental Design without Likelihoods by Desi R. Ivanova, Adam Foster, Steven Kleinegesse, Michael U. Gutmann, Tom Rainforth
    • Poster session: Wed 8 Dec 12:30 a.m. PST — 2 a.m. PST
  • On Optimal Interpolation in Linear Regression by Eduard Oravkin, Patrick Rebeschini
    • Poster session: Wed 8 Dec 12:30 a.m. PST — 2 a.m. PST
  • Stability & Generalisation of Gradient Descent for Shallow Neural Networks without the Neural Tangent Kernel by Dominic Richards, Ilja Kuzborskij
    • Poster session: Wed 8 Dec 4:30 p.m. PST — 6 p.m. PST
  • NEO: Non Equilibrium Sampling on the Orbit of a Deterministic Transform by Achille Thin, Yazid Janati El Idrissi, Sylvain Le Corff, Charles Ollion, Eric Moulines, Arnaud Doucet, Alain Durmus, Christian P Robert
    • Poster session: Thu 9 Dec 12:30 a.m. PST — 2 a.m. PST
  • Time-independent Generalization Bounds for SGLD in Non-convex Settings by Tyler Farghly, Patrick Rebeschini
    • Poster session: Thu 9 Dec 12:30 a.m. PST — 2 a.m. PST
  • Powerpropagation: A sparsity inducing weight reparameterisation by Jonathan Schwarz, Sid M Jayakumar, Razvan Pascanu, Peter E Latham, Yee Whye Teh
    • Poster session: Thu 9 Dec 12:30 a.m. PST — 2 a.m. PST
  • Outcome-Driven Reinforcement Learning via Variational Inference by Tim G. J. Rudner, Vitchyr H. Pong, Rowan McAllister, Yarin Gal, Sergey Levine
    • Poster session: Thu 9 Dec 8:30 a.m. PST — 10 a.m. PST
  • Conformal Bayesian Computation by Edwin Fong, Chris Holmes
    • Poster session: Thu 9 Dec 8:30 a.m. PST — 10 a.m. PST
  • Distributed Machine Learning with Sparse Heterogeneous Data by Dominic Richards, Sahand N. Negahban, Patrick Rebeschini
    • Poster session: Thu 9 Dec 8:30 a.m. PST — 10 a.m. PST
  • Group Equivariant Subsampling by Jin Xu, Hyunjik Kim, Tom Rainforth, Yee Whye Teh
    • Poster session: Thu 9 Dec 8:30 a.m. PST — 10 a.m. PST
  • Diffusion Schrödinger Bridge with Applications to Score-Based Generative Modeling by Valentin De Bortoli, James Thornton, Jeremy Heng, Arnaud Doucet
    • Spotlight presentation: Thu 9 Dec 4:30 p.m. PST — 6 p.m. PST
  • Uniform Sampling over Episode Difficulty by Sébastien M. R. Arnold, Guneet S. Dhillon, Avinash Ravichandran, Stefano Soatto
    • Spotlight presentation: Thu 9 Dec 4:30 p.m. PST — 6 p.m. PST
  • On Locality of Local Explanation Models by Sahra Ghalebikesabi, Lucile Ter-Minassian, Karla Diaz-Ordaz, Chris Holmes
    • Poster session: Fri 10 Dec 8:30 a.m. PST — 10 a.m. PST
  • On Contrastive Representations of Stochastic Processes by Emile Mathieu, Adam Foster, Yee Whye Teh
    • Poster session: Fri 10 Dec 8:30 a.m. PST — 10 a.m. PST
  • Implicit Regularization in Matrix Sensing via Mirror Descent by Fan Wu, Patrick Rebeschini
    • Poster session: Fri 10 Dec 8:30 a.m. PST — 10 a.m. PST
  • Multi-Facet Clustering Variational Autoencoders by Fabian Falck, Haoting Zhang, Matthew Willetts, George Nicholson, Christopher Yau, Chris Holmes
    • Poster session: Fri 10 Dec 8:30 a.m. PST — 10 a.m. PST

  We have a paper in the Datasets and Benchmarks Track:

  • Benchmarking Bayesian Deep Learning on Diabetic Retinopathy Detection Tasks by Neil Band, Tim G. J. Rudner, Qixuan Feng, Angelos Filos, Zachary Nado, Michael W. Dusenberry, Ghassen Jerfel, Dustin Tran, Yarin Gal
    • Poster at Datasets & Benchmarks Session
    • Spotlight at Workshop on Distribution Shifts
    • Poster at Symposium on Machine Learning for Health (ML4H)
    • Poster at Workshop on Bayesian Deep Learning

  We also have three workshop papers:

  • Uncertainty Baselines: Benchmarks for Uncertainty & Robustness in Deep Learning by Zachary Nado, Neil Band, Mark Collier, Josip Djolonga, Michael W. Dusenberry, Sebastian Farquhar, Qixuan Feng, Angelos Filos, Marton Havasi, Rodolphe Jenatton, Ghassen Jerfel, Jeremiah Zhe Liu, Zelda E Mariet, Jeremy Nixon, Shreyas Padhy, Jie Ren, Tim G. J. Rudner, Yeming Wen, Florian Wenzel, Kevin Patrick Murphy, D. Sculley, Balaji Lakshminarayanan, Jasper Snoek, Yarin Gal, Dustin Tran
    • Poster at Workshop on Bayesian Deep Learning
  • PCA Subspaces Are Not Always Optimal for Bayesian Learning by Alexandre Bense, Amir Joudaki, Tim G. J. Rudner, Vincent Fortuin
    • Poster at Workshop on Distribution Shifts
  • Uncertainty Quantification in End-to-End Implicit Neural Representations for Medical Imaging by Francisca Vasconcelos, Bobby He, Yee Whye Teh
    • Oral presentation at MedNeuRIPs Workshop
    • Poster at Bayesian Deep Learning Workshop

  And don’t miss Yee Whye Teh’s invited talk A Bayesian Perspective on Neural Processes at Bayesian Deep Learning Workshop on Dec 14!

• OxCSML at ICML 2021

  The group is participating in ICML 2021. Please feel free to stop by any of our poster sessions or presentations! We have 14 papers accepted to the main program of the conference:

  • Deep Adaptive Design: Amortizing Sequential Bayesian Experimental Design by Adam Foster, Desi R. Ivanova, Ilyas Malik and Tom Rainforth
  • Differentiable Particle Filtering via Entropy-Regularized Optimal Transport by Adrien Corenflos*, James Thornton*, George Deligiannidis, Arnaud Doucet
  • Improving Lossless Compression Rates via Monte Carlo Bits-Back Coding by Yangjun Ruan, Karen Ullrich, Daniel Severo, James Townsend, Ashish Khisti, Arnaud Doucet, Alireza Makhzani and Chris J. Maddison
  • Provably Strict Generalisation Benefit for Equivariant Models by Bryn Elesedy and Sheheryar Zaidi
  • Active Testing: Sample-Efficient Model Evaluation by Jannik Kossen, Sebastian Farquhar, Yarin Gal and Tom Rainforth
  • Probabilistic Programs with Stochastic Conditioning by David Tolpin, Yuan Zhou, Tom Rainforth and Hongseok Yang
  • Exploration in Approximate Hyper-State Space for Meta Reinforcement Learning by Luisa Zintgraf, Leo Feng, Cong Lu, Maximilian Igl, Kristian Hartikainen, Katja Hofmann and Shimon Whiteson
  • Augmented World Models Facilitate Zero-Shot Dynamics Generalization From a Single Offline Environment) by Philip J. Ball*, Cong Lu*, Jack Parker-Holder and Stephen Roberts
  • Think Global and Act Local: Bayesian Optimisation over High-Dimensional Categorical and Mixed Search Spaces by Xingchen Wan, Vu Nguyen, Huong Ha, Binxin Ru, Cong Lu, Michael A. Osborne
  • Monte Carlo Variational Auto-Encoders by Achille Thin, Nikita Kotelevskii, Alain Durmus, Maxim Panov, Eric Moulines , Arnaud Doucet
  • LieTransformer: Equivariant Self-Attention for Lie Groups by Michael Hutchinson*, Charline Le Lan*, Sheheryar Zaidi*, Emilien Dupont, Yee Whye Teh, Hyunjik Kim
  • Equivariant Learning of Stochastic Fields: Gaussian Processes and Steerable Conditional Neural Processes by Peter Holderrieth, Michael Hutchinson, Yee Whye Teh
  • Asymmetric Heavy Tails and Implicit Bias in Gaussian Noise Injections by Alexander Camuto, Xiaoyu Wang, Lingjiong Zhu, Chris Holmes, Mert Gürbüzbalaban and Umut Şimşekli
  • On Signal-to-Noise Ratio Issues in Variational Inference for Deep Gaussian Process by Tim Rudner, Oscar Key, Yarin Gal and Tom Rainforth

  In addition, Yee Whye Teh received the Test of Time Award for his 2011 paper with Max Welling

  • Bayesian Learning via Stochastic Gradient Langevin Dynamics

  See here for a quick run down of each paper, plus the presentations and poster sessions for each.

• 2 UAI 2020 Accepted Papers!

  2 papers co-authored by the OxCSML group members have been accepted to the main program of UAI 2020

• 8 ICML 2020 Accepted Papers!

  8 papers co-authored by the OxCSML group members have been accepted to the main program of ICML 2020

• 5 AISTATS 2020 Accepted Papers!

  5 papers co-authored by the OxCSML group members have been accepted to the main program of AISTATS 2020

• 2 ICLR 2020 Accepted Papers!

  2 papers co-authored by the OxCSML group members have been accepted to the main program of ICLR 2020

• DPhil opportunity on machine learning for climate science

  As a part of the iMIRACLI innovative training network funded by the European Union, there is a DPhil (PhD) opening at OxCSML to work on scalable and expressive spatio-temporal modelling for climate. This project will be supervised by Prof. Dino Sejdinovic and will be in collaboration with EPFL Lausanne, Amazon, and Met Office. Another 14 projects within the network are available. Candidates can be of any nationality but are required to undertake transnational mobility (i.e. move from one country to another) when taking up their appointment.

  Unfortunately, applications for this position have now closed.

• Yee Whye in I'm a Researcher: Machine Learning Zone

  OxCSML group’s Yee Whye Teh recently participated in I’m a Researcher: Machine Learning Zone, a fun two week online event connecting school students with academic researchers, allowing students to know bit more about the science behind machine learning as well as how life is like as a researcher in university. It is also a competition among researchers where students are the judges, and Yee Whye was the winner, congratulations!

• 19 Neurips 2019 Accepted Papers!

  19 papers co-authored by the OxCSML group members have been accepted to the main program of Neurips 2019

• OxCSML to take part in a new Innovative Training Network on machine learning and climate science

  The European Commission will fund an Innovative Training Network iMIRACLI (innovative MachIne leaRning to constrain Aerosol-cloud CLimate Impacts) which will train doctoral students in machine learning skills to address climate change.

  iMIRACLI will be led by the University of Oxford with the involvement of the Department of Physics (Philip Stier and Duncan Watson-Parris) and the Department of Statistics (Dino Sejdinovic). Network also includes the University of Leipzig, Stockholm University, ETH Zurich, the University of Edinburgh, Universitat de Valencia, University College London, the German Aerospace Center (DLR), Ecole Polytechnique Federale de Lausanne and the University of Jena, together with the non-academic partners Amazon, The Alan Turing Institute, MetOffice, Iris.ai, GAF AG and FastOpt.

  Further details are available here.

• Two papers by OxCSML members receive ICML 2019 Best Paper Honorable Mentions

  Two papers co-authored by the OxCSML members received honorable mentions in best paper awards at ICML 2019.

• 6 ICLR 2019 Accepted Papers

  6 papers co-authored by the OxCSML group members are being presented at the main program of ICLR 2019:

• 13 ICML 2019 Accepted Papers

  13 papers co-authored by the OxCSML group members have been accepted to the main program of ICML 2019:

• NeurIPS 2018 Workshop Participation

  Members of the group are organizing and participating in a number of NeurIPS 2018 Workshops and a co-located symposium.

• NeurIPS 2018 Accepted Papers

  13 papers co-authored by members of the group have been accepted to the main program of NeurIPS 2018.

• 4 UAI 2018 Accepted Papers

  4 papers co-authored by the OxCSML group members have been accepted to the main program of UAI 2018:

• 9 ICML 2018 Accepted Papers

  9 papers co-authored by the OxCSML group members have been accepted to the main program of ICML 2018:

• 3 AISTATS 2018 Accepted Papers

  3 papers co-authored by the OxCSML group members have been accepted to the main program of AISTATS 2018:

• Yee Whye's Breiman Lecture

  Slides and video for Yee Whye Teh’s Breiman keynote lecture can be found here.

• NIPS Workshops Participation 2017

  In addition to 6 papers in the main program of NIPS 2017 and Yee Whye Teh’s Breiman keynote lecture, OxCSML will be represented at several NIPS workshops with the following contributions.

• Postdoctoral Research Assistant in Statistical Machine Learning

  Applications are invited for a full-time postdoctoral research assistant in statistical machine learning, fixed-term for up to 2 years. Reporting to Professors Yee Whye Teh and Dino Sejdinovic, the postholder will be a member of the OxCSML (Oxford Computational Statistics and Machine Learning) research group with responsibility for carrying out research on the Oxford - Tencent AI collaborative project on Large-Scale Machine Learning. The funds supporting this research project are provided by Tencent AI until October 2020.

  Further details are here.

  The closing date for applications is 12.00 noon on Monday 8 January 2018. Interviews will be held on Friday 26 January 2018.

• Royal Society's Report on Machine Learning

  The Royal Society’s Machine Learning Working Group, which included Professor Peter Donnelly and Professor Yee Whye Teh, issued a report entitled Machine Learning: the power and promise of computers that learn by example.

  You can also hear about the report in the recent episode of the Talking Machines.

• Deep Learning Indaba 2017

  An Indaba is a Zulu word for gathering, meeting for the discussion of affairs of the community. Last September, such an Indaba took place in Johannesburg, South Africa under the title “the Deep Learning Indaba”. Its target was to attract students from South Africa and Africa in general and build an understanding of the principles and practice of modern machine learning. Dr Konstantina Palla, postdoctoral fellow in the OxCSML group, was invited to give a tutorial on probabilistic reasoning with a focus on its connections to deep neural networks.

• NIPS 2017

  Yee Whye Teh will give a Breiman keynote lecture at NIPS 2017 entitled On Bayesian Deep Learning and Deep Bayesian Learning.

  6 papers co-authored by the OxCSML group members have been accepted to the main program of NIPS 2017:

• Postdoctoral Research Assistant in Statistical Machine Learning

  Applications are invited for a full-time postdoctoral research assistant in statistical machine learning, to work on Bayesian nonparametric methods for recommender systems.

  Queries about the post should be addressed to Professor François Caron: caron@stats.ox.ac.uk.

  The closing date for applications is 12.00 noon on Friday 18 August 2017.

  Job details

• Diversity in Machine Learning

  On 25 May 2017 we will be hosting “Diversity in Machine Learning”, an event for undergraduates, featuring two great speakers, Stefanie Jegelka (MIT) and Raia Hadsell (DeepMind).

• "Sparse graphs using exchangeable random measures" by Caron and Fox: RSS discussion paper meeting

  “Sparse graphs using exchangeable random measures” by François Caron and Emily Fox will be presented to the Royal Statistical Society at the Discussion Meeting on Wednesday, May 10th at 5pm.

• AISTATS and ICLR 2017 Accepted Papers

  Three papers from the group have been accepted at AISTATS 2017 and one paper at ICLR 2017.

  The papers are:

  • Poisson intensity estimation with reproducing kernels by Seth Flaxman, Yee Whye Teh, Dino Sejdinovic

  • Relativistic Monte Carlo by Xiaoyu Lu, Valerio Perrone, Leonard Hasenclever, Yee Whye Teh, Sebastian Vollmer

  • Encrypted accelerated least squares regression by Pedro Esperança, Louis Aslett, Chris Holmes

  • The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables by Chris Maddison, Andriy Mnih, Yee Whye Teh

• "What is machine learning?" Animation Launched

  In collaboration with Oxford Sparks, machine learning group members Seth Flaxman, Hyunjik Kim, and Prof Yee Whye Teh created a two minute animation answering the question, “What is machine learning?”.

• NIPS 2016 participation

  Many group members will be at NIPS 2016 presenting work at the main conference and workshops.

  • Tamara Fernández will be presenting “Gaussian Processes for Survival Analysis” at the main conference.
  • Stefan Webb will be presenting “A Tighter Monte Carlo Objective with Renyi alpha-Divergence Measures” at the Bayesian Deep Learning workshop.
  • Hyunjik Kim will be presenting “Scalable Structure Discovery in Regression using Gaussian Processes” at the Practical Bayesian Nonparametrics workshop.
  • Leonard Hasenclaver, Stefan Webb and Thibaut Lienart will be presenting “Distributed Bayesian Learning with Stochastic Natural-gradient Expectation Propagation and the Posterior Server” at the Advances in Approximate Bayesian Inference and Bayesian Deep Learning workshops.
  • Valerio Perrone and Xiaoyu Lu will be presenting “Relativistic Monte Carlo” at the Bayesian Deep Learning workshop.
  • Konstantina Palla will be presenting “Bayesian nonparametrics for Sparse Dynamic Networks”, Xiaoyu Lu will be presenting “Tucker Gaussian Process for Regression and Collaborative Filtering”, Qinyi Zhang will be presenting “Large-Scale Kernel Methods for Independence Testing” and Jovana Mitrovic will be presenting “Disentangling the Factors of Variation at Initialization In Neural Networks” at the Women in Machine Learning Workshop.

• Postdoctoral Research Assistant in Machine Learning

  We will have an opening for a two-year full-time Postdoctoral Research Assistant, in the areas of kernel methods, Gaussian processes, or probabilistic programming. Queries should be addressed to Professor Yee Whye Teh (y.w.teh@stats.ox.ac.uk).

• Posterior Server Software Released

  We have released the source code implementing the Posterior Server. It can be found on GitHub at https://github.com/BigBayes/PosteriorServer. The code is written in Julia and is released under the MIT license.

• International Prize in Statistics awarded to David Cox

  Congratulations to Professor Sir David Cox on being awarded the first ever International Prize in Statistics in recognition of his many extraordinary contributions to statistics and science, especially his introduction of the proportional hazards model in a groundbreaking 1972 paper.

• Yee Whye Teh to Co-Chair ICML2017

  Yee Whye will be programme co-chair for ICML2017 along with Doina Precup, while Tony Jebara is general chair.

• DeepMind Scholarship

  We sincerely thank DeepMind for funding a DPhil Scholarship used to support Chris Maddison!

• Machine Learning Group Retreat

  We went for a group Summer retreat to Trogir, Croatia, 23-28 August. It was a great opportunity to update each other on current research projects, and a chance to discuss future projects and initiate collaborations in an informal, relaxed setting. Of course, it was much fun and great bonding time for the group!

• Postdoctoral Research Assistant in Machine Learning

  Applications are invited for two full-time ERC-funded Postdoctoral Research Assistants to work on the project ‘BigBayes: Rich, Structured and Efficient Learning of Big Bayesian Models’ in the Department of Statistics.