Institute of Communications and Connected Systems


Information-Theoretic Foundations of Data Processing: A Rate-Distortion Approach

Holding image - Awaiting image submission

1 September 2016

Using information-theoretic approaches to lay out the foundations of big data storage and processing systems.

Funder Royal Society
Amount £ 12 000

Project website https://www.ee.ucl.ac.uk/

Research theme logos - Sensing, Information and Data Processing
Research topics  Information Theory | Machine Learning | Data Analysis | Data Processing


The digital revolution – which has been underpinned by the emergence of digital computers, digital communications and more recently the Internet – has given rise to our current information age that is characterized by the generation, collection,analysis and processing of vasts amounts of data in various domains.

In science, modern scientific experiments and simulations in the medical & biological sciences, the physical sciences and engineering are on the verge of generating petabytes of data (1,000,000,000,000,000 bytes) and beyond.

In technology, there are billions of devices worldwide that constantly gather, generate, communicate and process data; also, Internet companies such as Google, Microsoft, and Yahoo currently boast zettabytes of data (1,000,000,000,000,000,000,000 bytes) and social media giants such as Facebook, Twitter and YouTube also boast hundreds of millions of users constantly generating and consuming data.

Business and commerce are also gradually capturing vasts amount of data associated with customer activities in order to spot business trends and opportunities.

In addition, the vision of a healthcare system that maintains increasingly detailed data for each individual—including genomic, cellular, clinical and environmental data—that can be combined with data from other individuals along with results from fundamental medical research in order to derive more effective personalised treatments is also being discussed worldwide.

This so-called big data – which offers the means to produce new insights – is generally acknowledged to lead to exciting opportunities. For example, the transition to data-rich healthcare systems offers the opportunity to introduce new technology that be used to improve health & well-being, prevent and cure disease or predict epidemics – a transformation that is extremely relevant to contribute to the sustainability of healthcare despite an increasingly ageing population andlimited resources in the UK and Europe.

However, big data is also confronting us with a challenge: it is often argued tha today’s ICT infrastructure – without radical changes to the way one collects, analyses and processes information – will not be able to keep up with the current pace of data growth in view of fundamental limitations in data transmission, data storage and computation capability.

Our research addresses this challenge by leveraging tools from the fields of information theory, information processing and optimization. In particular, our research asks the question: how can one capture, communicate and store data as efficiently as possible without missing relevant information? 

We answer this question by unveiling:

  • Fundamental limits in data acquisition & representation: These correspond to fundamental bottlenecks that cannot be overcome irrespectively of the sophistication and complexity of the data acquisition & representation methods;
  • Efficient schemes for data acquisition & representation: These schemes offer the means to capture, communicate and store data effectively without loosing relevant information.

Our research is very exciting because it offers us the opportunity to advance the foundations and applications of data processing but also – in view of its interdisciplinary nature – the opportunity to engage with colleagues from other fields. Our methods have the potential to influence the research agenda in our fields but also influence other sectors in the wider society.


None currently linked