Featured Publications

For full list, please refer to Google Scholar or DBLP

ICLR 2022

We show that it is possible to train graph neural networks that achieve state of the art accuracy, despite having memory usage and operation count proportional to the number of vertices in the graph–not the number of edges, as is the case with previous state of the art.
PDF Code Blog Post
ICLR 2022

We show that using meta-gradients–where we backpropagate through multiple training steps–to prune-at-initialization enables us to achieve state-of-the-art results on a variety of vision classification tasks, with less data and in a single shot compared to existing methods.
PDF
Deep Learning for Geometric Computing Workshop, ICCV 2021

We demonstrate that popular existing point cloud architectures have significant redundancies, that can be optimized away to create faster and more resource-efficient models. Awarded Best Paper at Deep Learning for Geometric Computing Workshop, ICCV 2021.
PDF
ICLR 2021 and ICML Graph Representation Learning and Beyond Workshop 2020

This work is the first studying quantization for graph neural networks. We show that it is possible to train networks that use low-precision integer arithmetic at inference time, resulting in substantially lower memory usage, while causing little or no loss in accuracy in many cases.
PDF Code
HotMobile 2020

In this paper we argue that activity recognition using inertial sensors, such as accelerometers, has stagnated. We make the case that imaging sensors should be the new default for activity recognition thanks to recent developments in the hardware and low-power machine learning field. Published at HotMobile ‘20.
PDF
Master's Dissertation

Explored the viability of capturing internal body sounds with a wearable device for medical applications. This project involved the construction of wearable device (with specialised acoustic circuitry) which was used to collect a dataset; the experimental procedure was designed to evaluate performance during challenging user activities and acoustic conditions. Algorithms for continuous heart monitoring and asthma symptom detection are described and evaluated. It was shown that the algorithms could be run in real-time on plausible wearable hardware.

Supervised by Professor Cecilia Mascolo. Awarded “Highly Commended” dissertation prize by the University. Work from dissertation published at WellComp workshop at UbiComp 2020, winning best paper.

PDF
Bachelor's Dissertation

Conceptually similar to the lost device tracking features included by Apple in iOS 13 (2019) and used by world governments for contact tracing. The dissertation describes a novel technique for Bluetooth 5 that enables privacy-preserving human-proximity detection. Supervised by Dr Robert Harle. Some results from dissertation included in a paper at PerCom 2018.
PDF

Industry Experience

 
 
 
 
 

Research Intern

ARM

May 2021 – Aug 2021 Cambridge, UK

Achieved 4.5x reductions in inference latency for models operating on point cloud data, without accuracy degradation. First author paper awarded best paper at the Deep Learning for Geometric Computing ICCV Workshop.

Worked with Partha Maji and Tiago Azevedo on the Machine Learning Research team.

 
 
 
 
 

Software Engineering Intern

Facebook

Jun 2018 – Sep 2018 London, UK
Interned on Workplace team in London. Designed and deployed to production REST APIs and bot features that are widely advertised as part of the core product offering.
 
 
 
 
 

Software Engineering Intern

Ensoft

Jul 2017 – Sep 2017 Harpenden, UK
Wrote an automated regression testing tool to verify that a system re-implementation matched the original system over billions of possible inputs.

Contact

  • sat62@cam.ac.uk
  • Department of Computer Science, 15 JJ Thomson Ave, Cambridge CB3 0FD