Outreach

The HiPEDS Centre is committed to providing outreach and public engagement activities. These activities often include partnerships with schools and relationships with industrial partners. The outreach component of the doctoral training programme allowed our students to develop skills to communicate their subject area more widely, as well as engaging the public with their science.

Our past students have created seven videos with Amyric Films that show some of the exciting areas of their research. We hope you’ll find them interesting!

Charlie Houseago teaches us about one of the most powerful tools available for computer vision: the convolutional filter.

How Robots See

Charlie Houseago teaches us about one of the most powerful tools available for computer vision: the convolutional filter.

Ali Hadian describes the challenges around how to work with big data so that it can be searched quickly.

Trouble with Big Data

Ali Hadian describes the challenges around how to work with big data so that it can be searched quickly.

Michael Akintunde describes the importance of verification of automated systems and how researchers are attempting to achieve this verification.

Can we Guarantee AI is Safe?

Michael Akintunde describes the importance of verification in automated systems and how researchers are attempting to achieve this verification.

Luca Castiglione discusses attacks on cyber physical systems, i.e. systems that interact with their surrounding environment. For example, a train network. How can we detect and defend against these attacks?

Security in Cyber Physical Systems

Luca Castiglione discusses attacks on cyber physical systems, such as systems that interact with their surrounding environment (e.g., a train network). How can we detect and defend against these attacks?

Most people find it quite easy to figure out who is talking, and when, just through hearing. However, computers find this very difficult.

Understanding the Spoken World

Aiden Hogg explains that, whilst most people find it quite easy to figure out who is talking, and when through hearing, computers find this very difficult.  He takes us through how computers do this now, and how his research allows computers to do this more easily.

Ruizhe Zhao takes us through his research, which aims to effectively reduce the size of these networks to improve the performance of these networks.

Deep Learning with Custom Devices

Ruizhe Zhao takes us through his research on how computers can use "Convolutional Neural Networks" to perform visual tasks, like detecting objects in videos. However, these can require lots of memory and be slow to execute. His research aims to effectively reduce the size of these networks to improve their performance.

Jessica Vandebon walks us through her research that aims to automatically apply optimisations for these heterogeneous systems to help programmers effectivly use these.

Metaprogrammes for Heterogenous Systems

Jessica Vandebon walks us through her research on enabling fast and energy efficient computing, where modern large-scale computer systems contain specialist hardware that can accelerate parts of an application. However, programming these devices currently requires specialist skills that few programmers do have.  Her research aims to automatically apply optimisations for these heterogeneous systems to help programmers use these effectively.

Kenny Malpartida Cárdenas talks about her research to develop tests for the detection antimicrobial resistance and infections diseases (including malaria and covid) that can be used without needing lab equipment.

Molecular-based point-of-care diagnostics

Kenny Malpartida Cárdenas talks about her research to develop tests for the detection antimicrobial resistance and infections diseases (including malaria and covid) that can be used without needing lab equipment.

Viet Pham Ngoc discusses his research of the use of Quantum Computers for machine learning.

Superpositition for Quantum Machine Learning

Viet Pham Ngoc discusses his research of the use of Quantum Computers for machine learning.

Convolutional Neural Networks (CNNs) are used in a wide variety of Machine Learning applications, but modern CNNs are limited by the time and energy that is needed to train them. Diederik Vink takes us through his research which aims to use custom solutio

Training of Machine Learning on Custom Hardware

Convolutional Neural Networks (CNNs) are used in a wide variety of Machine Learning applications, but modern CNNs are limited by the time and energy that is needed to train them. Diederik Vink takes us through his research which aims to use custom solutions on reconfigurable hardware to train CNNs. By calculating with less precise numbers where less accuracy is needed, CNNs can be trained fast and efficiently while maintaining accuracy!

Videos for the Public

A selection of HiPEDS student presentations and discussions of their research.

Films for the Public

Graphcore

Professor Andrew Davison and PhD researcher Joseph Ortiz discuss their use of Graphcore's IPU graph processor to more efficiently compute bundle adjustment in robot vision, using Gaussian Belief Propagation.

Joseph Ortiz

Graphcore

Joseph Ortiz discusses use of Graphcore's IPU graph processor

Professor Andrew Davison and PhD researcher Joseph Ortiz discuss their use of Graphcore's IPU graph processor to more efficiently compute bundle adjustment in robot vision, using Gaussian Belief Propagation.

Nicolas Perez Nieves presents his research at the online wor

Understanding the role of neural heterogeneity in learning

Nicolas Perez Nieves presents his research at the online workshop SNUFA'21.

Nicolas Perez Nieves presents his research at the online workshop: Spiking Neural Networks as Universal Function Approximators (SNUFA'21).

Kenny Malpartida describes digital diagnostic approaches

Lab on Chip to Detect Malaria

Kenny Malpartida describes digital diagnostic approaches

In this presentation “A lab-on-chip system to detect malaria”, Kenny Malpartida Cardenas (Department of Electrical and Electronic Engineering, Imperial College London) describes work bringing digital diagnostic approaches to malaria diagnosis.
 
FPL2020 live pitch for the full paper: Philippos Papaphilipp

FPL2020: (3-min) An Adaptable High-Throughput FPGA

Philippos Papaphilippo's live pitch for the full paper at FPL 2020

FPL2020 live pitch for the full paper: Philippos Papaphilippou, Chris Brooks and Wayne Luk "An Adaptable High-Throughput FPGA Merge Sorter for Accelerating Database Analytics", The International Conference on Field-Programmable Logic and Applications (FPL) 2020.

FPL2020 video presentation for the full paper: Philippos Pap

FPL2020: (25-min) An Adaptable High-Throughput FPGA

Philippos Papaphilippou's presentation for the full paper at FPL 2020

FPL2020 video presentation for the full paper: Philippos Papaphilippou, Chris Brooks and Wayne Luk "An Adaptable High-Throughput FPGA Merge Sorter for Accelerating Database Analytics", The International Conference on Field-Programmable Logic and Applications (FPL) 2020

Johannes Weiber explains how ROmodel extends the capabilitie

ROmodel: Modeling robust optimization problems in Pyomo

Johannes Weiber explains how ROmodel extends the capabilities of the modeling language Pyomo

Johannes Weiber explains how ROmodel extends the capabilities of the modeling language Pyomo to robust optimization problems.

A robust approach to warped Gaussian process-constrained opt

A robust approach to warped Gaussian process [...]

Johannes Weiber explains warped gaussian processes with robust optimization

Johannes Weiber explains how warped Gaussian processes combined with robust optimization can model black-box uncertainty in mathematical programs in an intuitive way - see the full research paper.
 
George Bisbas presents his research at the High Performance

Temporal blocking for wave propagation [...]

George Bisbas presents his research at the High Performance Computing Conference 2021, Rice Universi

George Bisbas presents his research at the High Performance Computing Conference 2021, Rice University, Ken Kennedy Institute, Houston.

Hongxiang Fan presents his research at ICCD'20.

Optimizing FPGA-Based CNN Accelerator [...]

Hongxiang Fan presents his research at ICCD'20.

Hongxiang Fan presents his research at ICCD'20.

Daniel Grumberg presents his research to the 2020 LLVM Devel

CompilerInvocation to -cc1 command line

Daniel Grumberg presents his research to the 2020 LLVM Developers' meeting.

Daniel Grumberg presents his research to the 2020 LLVM Developers' meeting.

Contact us

For general enquiries on how to work with the HiPEDS Centre, please get in touch with the Director of the HiPEDS Research Centre, Professor Wayne Luk

w.luk@imperial.ac.uk