James Newling |
06/2020 - present. Software technical lead. Graphcore. Bristol, UK. Graphcore has developed a new kind of processor for machine learning. I led a team developing PopART for a bit, now I work on Poprithms. These are 2 components of the Poplar software stack.
03/2018 - 05/2020. Machine learning frameworks developer. Graphcore. Bristol, UK. I worked on the initial development of PopART and its integration with ONNX and PyTorch.
01/2017 - 10/2017. Collaborator. Advanced Micro Devices (AMD) / HSA. Martigny, Switzerland. After my internship at AMD, I continued to collaborate on MIOpenGEMM . This was enabled through a grant from the HSA.
09/2016 - 12/2016. Co-op. Advanced Micro Devices (AMD). Austin, Texas. I was an intern in the software team developing ROCm. I started the MIOpenGEMM project for optimized GEMM (matrix multiply) written in OpenCL and C++. More information can be found on the project wiki.
09/2013 - 12/2017. Doctor of Philosophy (PhD). École polytechnique fédérale de Lausanne (EPFL) Lausanne, Switzerland. Under the supervision of François Fleuret. I obtained a PhD in computer science from the EDIC doctoral school at the École polytechnique fédérale de Lausanne. My thesis was in unsupervised machine learning. Source code for fast partitional clustering is available in the repositories zentas and eakmeans.
09/2013 - 12/2017. Research assistant. Idiap Research Institute. Martigny, Switzerland. I was a member of the Machine Learning Group at the Idiap Research Institute. Our group developed statistical learning techniques for computer vision, with a special focus on computational aspects.
03/2013 - 08/2013. Research assistant. The Mukherjee Lab for Statistical Systems Biology. Amsterdam, The Netherlands. I was a visitor in the Mukherjee Lab for Statistical Systems Biology, at the Netherlands Cancer Institute, where I worked on non-parameteric statistical tests of drug efficacy. Here is my final report (pdf).
09/2011 - 08/2013. Masters in Complexity Science École polytechnique, University of Warwick Paris, France and Warwick, UK. A description of the program is here. Some of the courses I took were numerical ODEs and SDEs, random models in evolution, theoretical neuroscience, modern statistical inference.
02/2006 - 10/2008, 02/2010 - 07/2011. Maths tutor and lecturer. University of Cape Town (UCT). Cape Town, South Africa. I tutored first year university mathematics at UCT, and I was later the course lecturer for a fourth year non-linear optimization course.
11/2009 - 07/2011. MSc in Applied Mathematics University of Cape Town (UCT) Cape Town, South Africa This was a research masters degree in observational cosmology, funded by the SKA. My thesis was awarded a UCT Research Associateship.
04/2009 - 09/2009. School maths teacher. Kathmandu University High Scool (KUHS). Dhulikhel, Nepal. I taught grade 6 maths at the Kathmandu University High School.
02/2004 - 11/2008. Honours degree, Pure Mathematics and Statistics. University of Cape Town (UCT). Cape Town, South Africa.
J. Newling and F. Fleuret. K-Medoids For K-Means Seeding. In Proceedings of the International Conference on Neural Information Processing Systems (NIPS) , pages 5201-5209, 2017. [arxiv] [bib] [poster] [slides for spotlight talk] [slides for SMLD talk]
J. Newling and F. Fleuret. A Sub-Quadratic Exact Medoid Algorithm. In Proceedings of the International Conference on Artificial Intelligence and Statistics (AISTATS) , pages 185–193, 2017. [arxiv] [bib] [poster] [Best paper award!]
J. Newling and F. Fleuret. Nested Mini-batch K-Means. In Proceedings of the International Conference on Neural Information Processing Systems (NIPS) , pages 1352–1360, 2016. [arxiv] [bib]
J. Newling and F. Fleuret. Fast K-means with Accurate Bounds. In Proceedings of the International Conference on Machine Learning (ICML) , pages 936–944, 2016. [arxiv] [bib] [poster] [slides]
Michelle Knights et al. Extending BEAMS to incorporate correlated systematic uncertainties. In Journal of Cosmology and Astrophysics. 2013. [arxiv]
James Newling et al. Parameter Estimation with BEAMS in the presence of biases and correlations. In Monthly Notices of the Royal Astronomical Society. 2012. [arxiv]
Renée Hlozek et al. Photometric Supernova Cosmology with BEAMS and SDSS-II. In The Astrophysical Journal. 2011. [arxiv]
James Newling et al. Statistical classification techniques for photometric supernova typing. In Monthly Notices of the Royal Astronomical Society. 2011. [arxiv]
R Kessler et al. Results from the supernova photometric classification challenge. In Publications of the Astronomical Society of the Pacific. 2010. [arxiv]
Thank you for visiting my webpage!