Computer Laboratory

Ben Day

Picture of Ben Day

I'm a PhD student at the University of Cambridge Computer Laboratory and a member of the Artificial Intelligence and Computational Biology groups under the supervision of Prof. Pietro Lio'. I previously worked in the group whilst studying for my Masters in the Department of Physics. I also tweet @itsmebenday.

My research interests are in novel Deep Learning architectures, particularly those that use recursive conditioning to enable self-understanding in learning systems and the application of graph based models to physics problems & non-Euclidean deep learning generally. I also work on taking a more scientific approach to building a 'theory of deep learning'. I supervise Masters student's research projects in both Computer Science and Physics and teach undergraduates. I'm happy to receive emails from students looking for a project supervisor.

When I'm not working I like to cook and garden. I can (technically) juggle three four balls and working on four five!

Recent highlights

  • 6/20 Our paper Uncertainty in Neural Relational Inference Trajectory Reconstruction has been accepted for presentation at the ICML workshop GRL+ and its fraternal twin Uncertainty in Multi-Interaction Trajectory Reconstruction has been accepted at the ICML workshop UDL
  • 4/20 Poster presentation at the ELLIS workshop on Geometric and Relational Deep Learning
  • 2/20 Attended AAAI 2020 in New York, presenting our work Proximal Distilled Evolutionary Reinforcement Learning with Cristian Bodnar
  • 12/19 Outstanding reviewer award (1 of 3) at the Graph Representation Learning workshop, NeurIPS 2019.

Publications and pre-prints

My Google Scholar is likely to be a more up to date resource.

  • Alexander Norcliffe, Cristian Bodar, Ben Day , Nikola Simidjievski & Pietro Liò (2020) On Second Order Behaviour in Augmented Neural ODEs. Pre-print. [arXiv]
  • Luís F. Simões, Ben Day, Vinutha M. Shreenath, Callum Wilson, Chris Bridges, Sylvester Kaczmarek, Yarin Gal (2019) FDL: Mission Support Challenge. Poster presentation at CiML workshop, NeurIPS 2019. [arXiv]
  • Cristian Bodnar, Ben Day & Pietro Liò (2019) Proximal Distilled Evolutionary Reinforcement Learning. Accepted and presented at AAAI'20. [arXiv]
  • Ezra Webb, Ben Day , Helena Andres-Terre & Pietro Liò (2019) Factorised Neural Relational Inference for Multi-Interaction Systems . Poster presented at the Learning and Reasoning with Graph-Structured Representations workshop at ICML 2019. [arXiv, poster]
  • Enxhell Luzhnica, Ben Day & Pietro Liò (2019) On Graph Classification Networks, Datasets and Baselines. Accepted for oral presentation (1 of 3) at the Learning and Reasoning with Graph-Structured Representations workshop at ICML 2019. [arXiv, poster, talk, slides]
  • Enxhell Luzhnica, Ben Day & Pietro Liò (2019) Clique pooling for graph classification . Poster presented at the Representation Learning on Graphs and Manifolds workshop at ICLR 2019. [arXiv, poster]
  • Conor Sheehan, Ben Day & Pietro Liò (2018) Introducing Curvature to the Label Space [arXiv]

Talks

Research Project Supervision

Part III Physics (Natural Sciences) Research Projects

Co-supervised with Prof Liò

  • Conor Sheehan (2018). Introducing Curvature to the Label Space . Project mark 78/100.
  • James Bernardi (2019). Physically Motivated Label Smoothing to Encode Human-Induced Segmentation Labelling Error
  • Jan Elsner (2019). Encoding Dynamics from Visual Data . Project mark 77/100
  • Ezra Webb (2019). Factorised Neural Relational Inference . Project mark 83/100 Tessella Prize for Best Computational Project
  • Chris Underhill (2020). Visualising the Loss Landscapes of Randomly Wired Neural Networks.
  • Alexander Norcliffe (2020). Second Order Neural ODEs .
  • Vasileios Karavias (2020). Uncertainty in neural relational inference .
  • Vijja Wichitwechkarn (2020). Multi-Interaction Systems: Interaction Analysis, Model Testing and Prioritised Sampling .

MPhil Computer Science Research Projects

Co-supervised with Prof Liò

  • Michelle Zheng (2018). . Machine Learning for Image Analysis of 3D Super-Resolution Microscopy. Project mark 82/100
  • Enxhell Luzhnica (2019). Best Project Prize
  • Cristian Bodnar (2019). Best Student Prize (Project & Coursework)

Teaching

  • Supervised Machine Learning with Real World Data Lent (2019,2020)
  • Supervised Foundations of Data Science Michaelmas 2019

Contact

Ben Day
Department of Computer Science and Technology
University of Cambridge
15 JJ Thomson Avenue
Cambridge CB3 0FD

Last updated 2019/09/14