FindAPhD Weekly PhD Newsletter | JOIN NOW FindAPhD Weekly PhD Newsletter | JOIN NOW

PhD Studentship: Self-supervised neuromorphic vision


   School of Engineering and Informatics

This project is no longer listed on FindAPhD.com and may not be available.

Click here to search FindAPhD.com for PhD studentship opportunities
  Dr James Knight  No more applications being accepted  Funded PhD Project (Students Worldwide)

About the Project

In this project you will combine SNNs and self-supervised learning to solve real-world tasks using event-based cameras.

Much of the recent deep learning revolution in computer vision is based on Artificial Neural Networks (ANNs) trained using supervised learning on fully labelled data. However, the computational cost of scaling these approaches to larger models has led many researchers to look to biology for answers. Spiking Neural Networks (SNNs) are inspired by the power efficient, sparse communication of biological neurons and recent advances [1,2] have allowed SNNs to achieve competitive performance on many machine learning tasks. However, they are beginning to encounter the same issue that hinders progress in ML more generally — obtaining labelled datasets large enough to train big models. Self-supervised approaches such as contrastive learning offer an exciting alternative and the performance of these approaches is now beginning to compete with supervised learning [3]. Furthermore, self-supervised learning has been shown to result in learned representations similar to those found in the activity of the visual cortex in smaller mammalian brains [4] and can be implemented using purely local learning rules [5].

In this project you will combine SNNs and self-supervised learning to solve real-world tasks using event-based cameras [6]. These cameras emit spikes rather than frames, making them ideally suited to providing input to SNNs and have a larger dynamic range and higher temporal accuracy than conventional frame-based cameras. To get up to speed quickly with these devices, this project includes a fully funded internship with event-based camera experts at the Institute of Neuroinformatics in Zurich and their iniVation spinout company. Building on this and using our own GPU accelerated SNN simulation framework [7], you will investigate how self-supervised learning can be applied to event based vision problems ranging from navigation to saccadic movements in image processing with the precise application being driven by your interests.

If you have experience with Machine Learning and programming in Python; and are interested in neuromorphic vision and SNNs we would love to hear from you.

  1. Wunderlich & Pehle (2021), Event-based backpropagation can compute exact gradients for spiking neural networks, https://doi.org/10.1038/s41598-021-91786-z
  2. Zenke & Neftci (2021), Brain-Inspired Learning on Neuromorphic Substrates, https://doi.org/10.1109/JPROC.2020.3045625
  3. Chen et al. (2020), A simple framework for contrastive learning of visual representations, http://proceedings.mlr.press/v119/chen20j.html
  4. Nayebi et al. (2021). Shallow Unsupervised Models Best Predict Neural Responses in Mouse Visual Cortex.  https://doi.org/10.1101/2021.06.16.448730
  5. Illinger et al. (2020), Local plasticity rules can learn deep representations using self-supervised contrastive predictions, http://arxiv.org/abs/2010.08262
  6. Lichtsteiner et al. (2008), A 128✕128 120 dB 15 µs Latency Asynchronous Temporal Contrast Vision Sensor. https://doi.org/10.1109/JSSC.2007.914337
  7. Knight et al. (2021), PyGeNN: A Python Library for GPU-Enhanced Neural Networks, https://doi.org/10.3389/fninf.2021.659005

How to Apply

Apply online for a full time PhD in Informatics using our step by step guide (http://www.sussex.ac.uk/study/phd/apply). Here you will also find details of our entry requirements.

Please clearly state on your application form that you are applying for the Self-supervised neuromorphic vision Scholarship under the supervision of Dr James Knight.

Contact us

For questions regarding the research project, please contact Dr James Knight ([Email Address Removed]).

Timeline

Application deadline: 23 February 

Interview date: March 2022

Notification date: early April 2022


Funding Notes

You will receive a tax free stipend at a standard rate of £15,609 per year for 3.5 years. Fees will be waived for 3.5 years for UK and exceptional international candidates. In addition, £9612 will be available to fund the internship in Zurich and the purchase of a high-end GPU workstation.
The stipend is available to: UK / EU / Overseas.
The fee waiver is available to: UK
Overseas applications are welcome, but overseas fees are only covered for exceptional candidates, applicants may need to fund the difference between the UK fees offered and the overseas amount.
Search Suggestions
Search suggestions

Based on your current searches we recommend the following search filters.

PhD saved successfully
View saved PhDs