Annual Conference 2005

1
Annual Conference 2005
2
Professor Dame Julia Higgins

• Chair, Engineering Physical Sciences Research
  Council

3
Programme

• 10am - Part 1 EPSRC Showcase
• 11am - Part 2 EPSRC Focus
• 11.30am Open Forum QA
• 12.45pm Lunch
• 2.30pm Part 3 the EPSRC Debate
• 4.30pm Afternoon tea and close

4
Showcase Interdisciplinary Research
Collaborations in ICT

• Special action taken in 1999
• Importance of Information and Communications
  Technologies (ITC) to the UK economy
• Actively encourage interdisciplinary working
• Stretch research boundaries
• Encourage critical mass

5
Showcase

• Call for outline proposals to establish
  Interdisciplinary Research Collaborations in ICT
• 101 submissions received
• 12 were selected and invited to submit full
  proposals
• In February 2000, five were selected to receive
  funding totalling 50 million over six years

6
Advanced Knowledge Technologies

• The Open University
• University of Aberdeen
• University of Edinburgh
• University of Sheffield
• University of Southampton
• Area information and knowledge management.

7
Dependability of Computer-based Systems (DIRC)

• City University
• Lancaster University
• University of Edinburgh
• University of Newcastle upon Tyne
• University of York
• Area Dependability of computer-based systems

8
Equator

• Royal College of Art
• University College London
• University of Bristol
• University of Glasgow
• University of Lancaster
• University of Nottingham
• University of Southampton
• University of Sussex
• Area Interweaving physical and digital
  interaction

9
Medical Images and Signals to Clinical
Information (MIAS)

• Imperial College London
• Kings College London
• University College London
• University of Manchester
• University of Oxford
• Area Medical imaging systems

10
Ultra-fast Photonics (UPC)

• Heriot-Watt University
• Imperial College London
• University of Bristol
• University of Cambridge
• University of Essex
• University of Glasgow
• University of St Andrews
• Area data communications

11
ICT Interdisciplinary Research Collaborations

• 5 Collaborations
• Research groups from 23 Universities
• Diverse disciplines
• Diverse research areas
• Diverse aspirations
• 50 million funding
• What has been achieved so far?

12
Interdisciplinary Research Collaborations An
Industry Assessment

• Andrew Herbert
• Managing Director
• Microsoft Research, Cambridge

13
IRC a New Model for Research
Social sciences
EQUATOR
Information sciences
Organisational sciences systems sciences
Core Computer Science
DIRC
AKT
e-Sciences
UFP
MIAS
Materials, physical sciences
Applications
Medical sciences
14
The Value of Interdisciplinary Research

• Enables challenging problems to be solved
• link teams complementary objectives, insights,
  technologies
• Innovation increasingly is located at boundaries
  between disciplines
• Enables end-to-end research linking theory,
  technology and applications
• stepping past traditional barriers
• push and pull 1 1 3
• ability to take on big problems

15
Example Dependability IRC

• Dependability requires an understanding of themes
  like risk (actual vs. perceived)
• Dependability failures often arise from
  mismatched mental models between system users and
  the state of the system
• Links computer scientists, sociologists,
  statisticians and psychologists to understand
  such issues
• Influencing how others will develop systems

16
One Example of a DIRC Study

• Advisory systems widely used, e.g. air traffic
  control
• Medical advisory systems also widespread
• DIRC sociologists studied reading of mammograms
• two humans (diversity if done properly) vs.
  Computer Aided Detection (CAD)
• prompt human to consider more closely
• One dependability study showed impact over all
  readers, DIRC statisticians studied impact on
  differentiated readers (and cases)

17
Effect of CAD on Probability of Detecting Cancer
(all cases)
The more sensitive readers hindered
Maximum improvement
0.85
0.80
Fraction of cases recalled by the reader without
CAD (sensitivity)
0.75
0.70
The less sensitive readers benefit
0.65
Maximum hindrance
0.60
0.0
0.2
0.4
0.6
0.8
More of the difficult cases missed
More of the easy cases detected
Fraction of readers missing the case without CAD
(case difficulty)
18
Taking on Big Problems

• Six year time scale allows for risk taking and
  building infrastructure
• Example MIAS grand challenges
• multi-scale modelling - e.g., cerebral
  circulation
• structure function e.g., combine brain models
  and feature recognition to register group brain
  images
• intelligent acquisition e.g., removing motion
  effects from MRI images, attenuation correction
  for in vivo ultrasound scan

19
Tilt testing at the Oxford Falls Clinic
As the tilt, and rate of tilt, are changed the
subjects blood pressure and heart rate vary
markedly. Autoregulation of cerebral blood flow
is a complex process (as the control model on the
next slide indicates). We have discovered that a
parameter called Tissue Oxygenation Index is key
to controlling autoregulation.
20
Some model control paths
21
Cerebral Circulationbiophysics biochemistry
signal processing milliseconds to hours

Syncope (fainting) patient data in the Oxford
Falls Clinic, remains constant for most of a
Tilt Test, though blood pressure decreases
markedly as the test proceeds. This patient does
not show any impairment of the auto-regulation
process
The left-hand side above shows a novel model
developed in the IRC predicting that the Tissue
Oxygenation Index (TOI) will remain constant as
blood pressure changes within 40 of its mean
value, demonstrating the phenomenon of cerebral
auto-regulation.
22
Achieving End-to-End Research

• Example Ultrafast Photonics
• femtosecond lasers advanced sources of
  ultrashort pulses
• photonic bandgaps miniaturization of optical
  circuitry
• inorganic semiconductors quantum dots
• organic materials plastics for data
  communication
• advanced systems and networks e.g., coding
  schemes, theory and modelling

23
Femtosecond lasers for datacomms
Step 1 The development of a novel ultrafast
laser designed to meet stringent system
requirements
24
fs-Lasers for data communications
Step 2 Implementation in systems
25
fs-Lasers for data communications
Step 3 World beating results!
1.36 Tb/s
Step 4 A new generation!
26
Benefits of Collaborative Research

• Multi-site projects with lightweight management
  structure
• Bringing together the best in the UK
• IRCs extended existing bi-lateral relationships
• Resources harness critical mass, retain local
  autonomy
• 40-60 researchers in each IRC
• Break free of institutional limits
• Greater responsiveness in interdisciplinary areas
  -already exploiting synergies between IRCs,
  especially for e-Science programme

27
Internationally Competitive Research
Prix Ars Electronica Golden Nica, ACM DIS design
award
28
IRCs Have Broader Impact

• AKT
• 20 representations to standards bodies (W2C)
• AKTive Space, first prize in Semantic Web Grand
  Challenge
• Tim Berners-Lee visiting Professor
• Equator
• UK and international press and TV
• Strong links to international industry research
  (Microsoft, Intel, PARC, HP, BT Exact)
• Both formative role in UK Grand Challenges and
  e-Science programme, EU Framework Programme

29
IRCs and Industry Engagement

• Challenges
• Industry cant commit to long time scales
• Impact of MA on partners and markets
• Yet each IRC has good industry engagement
• Bandwagons more successful than arranged
  marriages
• Entrepreneurial partnerships between academics
  and start-ups and established industry for
  example Oxford Biosignals

30
Example AKTORs Club

• Abbey National, Adiuri Systems, Aerosystems
  International, Airbus UK, Active Navigation,
  Applied Knowledge Management, Applied
  Intelligence, BAE Systems, Baker Hughes, BG
  Technology, Boeing, British Airways, British
  Maritime Technology, BP Sunbury, BP Amoco,
  British Telecom, Cacheon, Cancer Research UK, The
  Cape Alliance, CCLRC Rutherford Appleton
  Laboratory, Cedar Group, Clifford Chance, CQR
  Data Limited, Dstl, East of England Development
  Agency, Electronic Publishing Service, Emorphia,
  Empolis, Epistemics, European Office of Aerospace
  Research Development, Factiva, Fujitsu,
  Hewlett-Packard, Honeywell Hi-Spec Solutions,
  IBM, Logica UK Limited, Magnetical, McKinsey
  Company, Mitre, Napp Pharmaceuticals, Nature,
  Open Knowledge Network, Parametric Technology
  Corporation, PJR Consultants, PriceWaterhouseCoope
  rs, QinetiQ, Reuters, Rolls-Royce, Royal
  Institute of International Affairs, Solcara,
  Stilo Technology, Sun Microsystems, TFPL Ltd,
  Unilever, Vivas Limited, WebSymposia

31
Human Capital and Training

• IRC directors experienced, entrepreneurial
  research leaders - long term funding enables risk
  taking
• Career development for young researchers
• A new breed of PhD students
• working on challenging problems
• breadth as well as depth in research
• skilled in team work
• between groups at home institution
• between partner institutions
• MIAS Doctoral Training Scheme
• large numbers e.g., Equator has 40 PhD students
• proving attractive to UK graduates

32
Issues Going Forward

• Building an effective IRC takes time
• people need to learn to listen, build trust
• Balance between core computer science and
  applications - IRCs depend on ongoing research in
  core computer science
• Cant do everything where to focus?
• Bigger asks from IRC follow-on projects
  1-1.5M project costs before FEC
• Referees and review process for these and future
  IRCs

33
In Summary

• The experiment worked!
• EPSRC has the opportunity to sustain world
  beating research
• tackling big challenges
• harnessing critical mass research resources
• being more responsive in interdisciplinary areas
• Collaboration is effective, thanks to modern
  communications technology
• Sustain the momentum
• exploit synergies between IRCs
• boundaries and topics can change over time

34
EPSRC Showcase

• Open Forum QA