|
Affective Computing
Professor Ruth Aylett, Heriot-Watt University
Affective computing is a new research area that concerns the
development of systems that can register, model and/or influence human
emotional and emotion-related states and processes. It is an
inter-disciplinary area involving researchers from signal processing,
psychology, interface design, animation and a number of AI fields such
as agent architectures, robotics and reasoning. AI researchers have been
attracted to the area because of the recognition that emotion in the
broad sense pervades human communication and cognition. Human beings
have positive or negative feelings about most things, people, events and
symbols. These feelings strongly influence the way they attend, behave,
plan, learn and select. The feelings are conveyed e.g. in faces, voices,
gestures, and postures; and people judge others by the way they respond
to such signals. This talk gives an overview of significant work in
affective computing focusing on the role of emotion in autonomous agent
action-selection systems. The FearNot! interactive drama system in which
emotion-driven autonomous graphical characters are used for education
against bullying will be used as an example of an affective system.
Biography
Ruth Aylett is Professor of Computer Sciences in the School of
Maths and Computer Science at Heriot-Watt University. She has a degree
in mathematical economics from the LSE, and entered ICL as a graduate
trainee in 1976. After three years of technical support she moved to
Sheffield University where she worked in the micro-computing laboratory
and became interested in Artificial Intelligence. Whilst a lecturer at
the Sheffield Hallam University, she developed her interests with
particular reference to cognitive modelling, natural language and
intelligent interfaces. In 1989 she took up a post at the Artificial
Intelligence Applications Institute in Edinburgh University,
specialising in knowledge acquisition and knowledge engineering
methodologies, followed by two years as leader of the AI group at the
National Advanced Robotics Research Centre at Salford University. In
1998 she moved to the newly set up Centre for Virtual Environments and
located her research in the overlap between 3D interactive graphics and
artificial intelligence, first as Senior Lecturer, and then, from 2000,
as Professor of Intelligent Virtual Environments. She moved to
Heriot-Watt University in 2004.
Ruth's web pages
Directed Intervention Crossover applied to Bio-Control
Scheduling
Dr Julie Cowie, University of Stirling
In this talk I will discuss my interest in optimisation methods,
specifically focusing on current research examining the use of novel
techniques for optimising scheduling problems. Two novel Genetic
Algorithm cross-over techniques have been developed: Calculated
Expanding Bin (CalEB) and Targeted Intervention with Stochastic
Selection (TInSSel), which actively choose an intervention level and
spread of interventions based on the fitness of the parent solutions
selected for crossover.
CalEB and TInSSel are currently being evaluated by applying the
techniques to two diverse application areas: a bio-control problem and
chemotherapy treatment scheduling. The bio-control problem focuses on
the area of mushroom farming, where the farmer seeks to optimise the use
of the nematode Steinernema feltiae as a bio-control agent to control
sciarid flies. Issues surrounding the application of the nematode
include dosage levels and frequency of dose. In chemotherapy treatment,
the aim is to maximise the effect of the chemotherapy treatment, whilst
minimising the number of doses required.
Experiments to date have focused on the application of the techniques to
the bio-control problem. Results indicate that the CalEB and TInSSel
approaches lead to significant improvements over more traditional forms
of genetic algorithm (such as uniform crossover) when a penalty is
introduced for each intervention point used by the crossover algorithm.
The basics of genetic algorithms, the two approaches developed and the
results to date will be discussed in the talk.
Biography
Julie Cowie is a lecturer in the department of Computer Science
at Stirling University.
Her research interests include the use of Intelligent Decision Support
Systems (DSS) in the health service, with reference to health logistics.
This concerns the day to day running of health care establishments and
how the use of DSS and data modelling tools and techniques play a role
in this area. She is also interested in the role of decision support
systems in decision making. More recently, Julie has been looking at the
use of optimisation techniques such as genetic algorithms and the use of
neural techniques. This interest has lead her to investigate ways in
which different operational research and artificial intelligence
techniques might be combined to provide enhanced decision support. She
has been involved in applying her ideas in diverse areas, such as
optimising intervention points for bio-control agents in mushroom
farming, the application of soil science techniques to archaeological
questions, and providing an integrated model for early diagnosis of
dementia. From 1996 to 2000 she conducted research at the University of
Strathclyde towards her Ph.D. She is a key member of the
multi-disciplinary Medical Informatics research group, which spans the
Departments of Computing Science and Mathematics, Management and
Organisation, Nursing and Midwifery, and Psychology.
Julie's web pages
Intelligent Formulation Software: Theory, Memory or
Fantasy?
Professor Susan Craw, The Robert Gordon University
Chemical Formulation is important for the development of
products in a range of industries: pharmaceuticals, ceramics, plastics,
... even Formula 1 racing! It involves component-based design in which
the physical properties of the components deliver the required
properties of the product and chemical properties determine the ability
to form
the product. Balancing these requirements ensures that formulation is a
demanding problem-solving task. Formulation is the first step in
developing many products that involve the design and control of a
complete manufacturing process.
This talk seeks to answer the questions in the title. Is the
intelligence of formulation software a reality or just "in theory"? Can
memories invoke intelligence? Or is intelligent formulation software a
fantasy? It explores knowledge-based approaches behind intelligent
formulation software, the challenges in generating competitive
formulations and the opportunities offered by self-adapting decision
support tools. Will smart techniques enable theory or memory to achieve
the fantasy of intelligent formulation
software? Are there other applications on the horizon that may exploit
these intelligent technologies?
Biography
Susan Craw is Head of Research and Graduate Studies for Design &
Technology at the Robert Gordon University, Aberdeen. She graduated from
the University of Aberdeen with a 1st Class Honours degree in
Mathematics (1977) and a research MSc in Mathematics (1979) on "Homotopy
in Banach Algebras". Her first Computing qualification, a PhD on
"Automated Refinement for Knowledge Based Systems", came much later in
1991. As a Mathematics teacher at Mackie Academy, Stonehaven, she first
developed her interest in Computing, writing early software for
classroom use in mathematics. In 1983 she joined RGIT (now RGU) as a
lecturer, doing her part-time PhD, and then developing her research
interests in the use of Machine Learning in Automated Knowledge
Engineering Tools. Her 70+ Computing publications are complemented with
one substantial mathematics journal paper! In 1998 she was awarded the
title of Professor and became only the second female Professor of
Computing in Scotland! She was appointed Head of the School of Computing
in 2001 and Head of Research and Graduate Studies for Design &
Technology in 2006. Despite spending all her academic career at RGU, she
has the memorable experiences of sabbaticals at Kyoto University in
Japan and the University of California at Irvine.
Susan's web pages
|
|
