MODELLING and SCIENTIFIC COMPUTING

1
MODELLING and SCIENTIFIC COMPUTING

• Complex Systems associated Computational Models
  and Data Analysis

2
COMPUTATIONAL vs COMPUTER SCIENCE

• Challenges
• Exploration of models of the natural and
  artificial world, where complexity precludes
  exact solutions.
• Computational Approach third arm to theory
  and experiment.
• Complex Systems - highly diverse Features e.g.
  Many parameters, Large Volume of data and need to
  mine it, Many basic elements and events.

• Key Phases
• Development of the maths. model or data sourcing.
• Development / Implementation of the algorithm
  for numeric solution
• Generation of solutions by e.g. numerical
  simulation of the phenomenon of interest, data
  mining/ matching/ analysis.
• Visualisation / Interpretation and Validation

3
CURRENT Group Interests

• Modelling of Stock Markets/ Strategies for
  Financial Auditing, (M. Crane, G. Keogh, S.
  Sharifi / J. Horgan, Y. Bimpeh)
• Agricultural models / Employment patterns
  (L.Killen et al.)
• Automatic Spoken Language Identification by
  Genetic Programming (A. McLoughlin)
• Spatiotemporal Models in Physical, Bio- and
  related systems e.g. immune response, traffic
  flow and finance, cellular networks. Associated
  statistical models. (H. Ruskin, Y. Feng. R. Wang,
  J. Burns, B. Zhu and Y. Liu)
• Mathematical/ Statistical Models - recognition of
  complex, deformable shapes/ patterns e.g.
  application to sign-language recognition,
  patterns in financial markets. (A. Sutherland, H.
  Wu, A. Shamaie)
• Aerospace - control systems. Environmental
  Pollution (water quality) (L. Tuohey)

4
COLLABORATIONS

• INCLUDE
• Internal Schools (Physics, Chemistry and Science
  Ed.), DCUBS, DCU-
• based centres, e.g. CDVP also
  HEA PRTLI - centre Nat.
• Inst. for Cellular Biology.
  Sub-theme in DCU Strategic Plan.
• National TCD, (Depts. Maths., Physics and
  Economics), UCD, (Depts.
• Physics, Agr. Econ.), NUIM,
  Teagasc, Dell, Hitachi, CAPTEC,
• Regional Centre of Dermatology,
  Mater, St. James Hosp.
• International Porto Uni., EU consortium and COST
  initiative, US,
• (USM, Naval Research Lab
  (Entropics) and Naval PG School).

5
CONFERENCE TARGETS - last 5 years only

• INCLUDE - papers given at
• Europhysics Conf. on Computational Physics,
  International Conference on Computational
  Sciences, American Physical Society Annual
  Meeting, VECPAR, (Vector and Parallel Processing
  -biannual international conference), IASTED
  conferences on Modelling and Simulation, on
  Modelling, Identification and Control, IMACS on
  Scientific Computation, Applied Mathematics and
  Simulation, CASI. World Congress of Pharmacy
  Pharmaceutical Sciences/Int'l Congress of
  FIP,EUROSIM Congress/ (Modelling and Simulation
  in Biology, Medicine and Biomedical Engineering),
  Irish Machine Vision and Image Processing, IEEE
  Workshops on PR and JAVA Optimisation, SPIE
  International Symposium on Multispectral Image
  Processing and Pattern Recognition, IASEL (Irish
  Association of Science Education and Learning).

6
PUBLICATION TARGETS- last 5 years only

• INCLUDE - papers published in
• Proceedings - as conferences previously
• Journals - J. Stat. Phys., Physica A,
  Theor. in Biosciences, Computer Physics
  Communications, Lecture Notes in Comp. Science,
  Computers Education, J. of the South African
  Institute of Computer Scientists and Information
  Technology, J. Acc. And Business Res., J.
  Practice and Theory, Amer. Acc. Asoc., J. Bus.
  And Ec. Stats, JASA, ERCIM Journal, Computers and
  Electronics in Agriculture, Labour, J. of
  Software Quality, J. of the American Society for
  Information Systems and Technology.
• Other Reports - Technical in-house etc.
  

7
BRIEF EXAMPLES STOCK MARKETS (M. Crane, G.
Keogh, S. Sharifi)
The Issue Market Crashes. Crash detection using
eigenvalues and eigenvectors of correlation
matrices. Use of Random Matrix Theory to
identify noisy and non-noisy parts of correlation
matrix. Identification and tracking of market
movements. Approach Fractional Calculus in
Finance Use of models with fractional derivative
powers (previously used in memory processes).
Identification of volatile periods and their
signatures using FC Models Applications of FC to
financial correlation matrices to model time and
frequency effects.
8
BRIEF EXAMPLES STRATEGIES FOR FINANCIAL
AUDITING (J. Horgan, Y. Bimpeh)

• The IssueDeveloping sampling and estimation
  procedures for rare incidence, skewed accounting
  populations
• Need Classical statistical theory does not apply
• Approach Bayesian, non-parametric and computer
  intensive methods of estimation.

9
BRIEF EXAMPLES SOCIO-EONOMIC and AGRICULTURAL
STUDIES (L.Killen)
The Issue Non-standard Employment
Patterns Approach Comparative Cross-sectional
statistical analyses incorporating trends over
time. Need Methodologies for determining
Characteristic Lactation Curves in Dairy
cattle. Approach Multivariate Analyses of
constituent factors for general data sets,
incorporating milk constituents, breed info. etc.
10
BRIEF EXAMPLES AUTOMATIC SPOKEN LANGUAGE ID(A.
McLoughlin)

• The Problem Creation of Automatic
  Spoken-Language Identification Programme.
• The Issues No specialised linguistic expertise,
  No labour-intensive labelling of training data,
  addition of new languages without extensive
  re-training, large-vocabulary continuous speech
  recognition. Uses Telephone call routing,
  International enterprises, emergency services,
  law enforcement and intelligence, real-time
  systems.
• The Approach Evolutionary Computation, (modelled
  on biological evolution and mutation), Genetic
  Programming.

11
BRIEF EXAMPLES FUNDAMENTAL CELLULAR SYSTEMS
(H. Ruskin, Y. Feng, R. Wang, B. Zhu,
Y.Liu)

• The Problems Fundamental Spatiotemporal
  Processes e.g. complexity as for disordered
  cellular networks (froth coarsening), many
  cell-types (models of human immune-response,
  vehicle units in traffic flow.
• Principal Issues Quantification of
• behaviour. Prediction.
• Approaches Microscopic Models,
• statistical estimation/physical laws, robust
• simulations. Links Bioinformatics

12
BRIEF EXAMPLES Models in recognition of complex,
deformable shapes/ patterns (A. Sutherland, H.
Wu, A. Shamaie)

• The Issues Nature of Model Choice - definition
  of shapes, boundaries distributions of shapes
  texture and feature extraction types of
  classifier, etc.
• Applications Pattern matching in finance, hand
  and other-gesture recognition - (best models and
  estimates), ecological competition, security
  systems and so on.
• Approach Mechanics may include real-time video
  need to define spatio-temporal process and
  template matching, or e,g, generate training data
  for statistical models, such as the Discrete
  Hidden Markov Model.

13
BRIEF EXAMPLESAEROSPACE CONTROL SYSTEMS /
ENVIRONMENTAL SCIENCE (L. Tuohey)

• Experience Several European Space Agency
  missions (development of on-board spacecraft
  control SW, payload calibration). Quality aspects
  of comms. SW for civil aviation. Software
  Engineering processes and quality.
• Current Concerns Modelling, simulation, and
  numerical techniques in estimation of (lake)
  water quality via remote sensing. Aim viable
  model of water quality from optical measurement.
• Approach Solution of so-called inverse
  problem.
• Plans Application to related problems in
  geophysics

14
THE FUTURENEW PROJECTS and DIRECTIONS/ The
FUNDING?

• PAST FUNDING EXPERIENCE EI Basic (submitted
  since 1997), ATRP(2001), RIF (2002), SFI
  (2001), other minor EI schemes, Teagasc - few
  specific targets however - record variable
• WHY? - ?? Quality of Research ?? (see pub.
  records and of good Impact Factor conferences
  and journals).
• SCHEMES - FEW for area to date large-scale
  less appropriate, joint/non-specialist panels not
  seen as favouring inter-disciplinary work unless
  specified areas. Inter-disciplinarity ? fewer
  beans typically? Non-networkers? Product - not
  immediately commercial.
• REMEDY- target specialists - (e.g. Teagasc 2002
  -success current discussions with Goldman-Sachs,
  EU consortium on COST and Asia-Link - more
  favourable focus, ?? OTHERS ?? - to be explored.