Community Research and Development Information Service - CORDIS

Periodic Report Summary 1 - SPARTAN (Sparse Representations and Compressed Sensing Training Network)

The SpaRTaN FP7 Marie Curie Initial Training Network is training a new generation of interdisciplinary researchers in sparse representations and compressed sensing, contributing to Europe’s leading role in scientific innovation.
By bringing together leading academic and industry groups with expertise in sparse representations, compressed sensing, machine learning and optimisation, and with an interest in applications such as hyperspectral imaging, audio signal processing and video analytics, this project is creating an interdisciplinary, trans-national and inter-sectorial training network to enhance mobility and training of researchers in this area.
Our research aim is to investigate new methods for sparse representations and compressed sensing, and to apply these new methods in key application areas. To achieve this, we have the following research objectives:
* To investigate new methods for sparse representations, including analysis sparsity, structural information, and dictionary learning;
* To use the developed methods for the analysis of a range of signals, such as:
- magnetic resonance imaging (MRI);
- hyperspectral imaging;
- audio and music signals;
- video sequences;
* To apply these methods to real-world problems, in particular through working with the private sector;
* To share research advances in theory and applications across the network;
* To disseminate the results of this research to a wide range of audiences.
SpaRTaN has recruited 8 university-based ESRs and 1 SME-based ER across 5 countries. The network has trained its Fellows in transferable skills, research software development and scientific topics.
Researcher Training
SpaRTaN held researcher development training in September 2015 and April 2016 to introduce ESRs to the challenges of crossing cultures, how to manage themselves and their supervisors, how to present their research and themselves, applying ethics to their research, their professional development.
Software Carpentry courses were run to introduce researchers to the software tools to help them do better research. With a focus on reproducibility and reusable code, the skills are designed to be useful in both academia or industry.
In April 2016 the project teamed up with the MacSeNet ITN to host a Spring School offering lectures and tutorials covering the theory of sparse representations, compressed sensing and related topics, alongside applications of these methods in areas such as image processing, audio signal processing, and signal processing on graphs.
The Spring School was open to researchers outside the network for free. This increased the number of attendees and the networking opportunities. All the Spring School course materials can be found on our website.
Scientific Results
Since they started our Fellows have developed:
*An exceptionally fast method for hyper spectral image denoising, achieving state of the art results[1]
*A solution to complex inverse problems, through conversion of the signal to a novel domain[2]
*A convergence acceleration technique for generic optimisation problems[3]
*A method for audio source separation using deep neural networks[4]
*A fast method for kernel dictionary learning using sparsity constraints for increased speed[5]
*A novel method for fast quantitative brain MRI with statistical modelling in multichannel coils (submitted)
*A state of the art image denoising method using class-specific image databases (submitted)
Lessons Learnt
*During the first 24 months of running an ITN many lessons have been learnt and a list is being compiled for others. These include:
*Allow sufficient time for recruitment – it takes longer than you expect.
*Recruiters and organisers need to be aware of visa needs for recruiting and arranging events.
*Allocate time to help ESRs/ERs to understand what they need to do as part of the network.
*Identify good local trainers in each partner institution to help build training programmes.
*Include visits to local teams during events, helping researchers to share their work with the network.
*Combine events for economies but be sure to include time for people to digest what they have learnt.
*Researcher Development Training works across fields and a mixed cohort is good for discussions.
*Internet is important – always check that the WiFi in your venue is ready for the size of your group.

Planned Work and Impacts
As well as continuing with state of the art research the Fellows will have another two soft skills training events, one in reproducible research at EPFL in November 2016 and another in science communication in Edinburgh in 2017. The latter will coincide with the International Science Festival, allowing ample opportunities for public engagement and science outreach. We are also running a Summer School in collaboration with the SPARS workshop in Lisbon in 2017 and we will open this up to international researchers to maximise the impact of the event. Through these events we will encourage our Fellows to develop into “T shaped People” capable of adapting in a fast-moving research field. As the Fellows start out on their secondments, many of them to SMEs, we will be encouraging them to experience the research industry and consider the impact their work can have in the real world.

Contact Details
SpaRTaN is funded under the FP7-PEOPLE-2013-ITN call and is part of the Marie Curie Actions — Initial Training Networks (ITN) funding scheme: Project number – 607290.
SpaRTaN is co-ordinated by Prof. Mark Plumbley at Surrey University.
The network Administrator is Dr. Helen Cooper.
Email enquiries regarding the project should be sent to:

[1] L. Zhuang, J. M. Bioucas-Dias, "Fast hyperspectral image denoising based on low rank and sparse representations", In: IEEE International Geoscience and Remote Sensing Symposium (IGARSS), 2016.
[2] R. Pena, X. Bresson, P. Vandergheynst, "Source localization on graphs via L1 recovery and spectral graph theory", In: IEEE 12th Image, Video, & Multidimensional Signal Processing Workshop, 2016.
[3] D. Scieur, A. d'Aspremont and F. Bach, “Regularized nonlinear acceleration” In: Advances in Neural Information Processing Systems, pp. 712-720, 2016
[4] A. Zermini, Y. Yu, Y. Xu, W. Wang and M. D. Plumbley, “Deep neural network based audio source separation” Accepted for: 11th IMA International Conference on Mathematics in Signal Processing, Birmingham, UK, 12-14 December 2016
[5] C. O'Brien and M. D. Plumbley, “Sparse kernel dictionary learning.” Accepted for: 11th IMA International Conference on Mathematics in Signal Processing, Birmingham, UK, 12-14 December 2016.

Reported by

United Kingdom


Life Sciences
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top