Past Seminars
Seminars held over the last six months.
- 12/06/2013: Dr. Ryo Kawasaki - Retinal Vascular Signs and Cardiovascular Diseases, and Beyond… -New Insights into Old Classics-
Abstract
Retina provides a unique window for us; we can literally observe microvasculature directly with simple non-invasive instruments. There is an article in 1859 already which reported a relationship between hypertension and retinal vascular signs. Since then, ophthalmologists and researchers have been linking retinal pathologies and systemic diseases, especially cardiovascular diseases such as hypertension and stroke.
In the last decade, quantitative retinal vascular measurements were introduced to investigate this link in detail and found that a subtle change in retinal vasculature may be considered as 'a surrogate marker' reflecting microcirculation in whole body. These quantitative assessments have been applied to large scale population-based epidemiological studies of both adults and children. These studies now convincingly show links between a variety of retinal microvascular signs to both clinical and subclinical systemic diseases including cardiovascular and metabolic diseases. All of these advancements were achieved in collaborative research of ophthalmologists and researchers in biomedical image analysis
This presentation will outline how traditional retinal vascular signs have been utilized in systemic cardiovascular disease assessment, and how modern retinal vascular assessments will contribute to improve systemic disease risk assessment.
BIO
Dr. Kawasaki is currently Assistant Professor of Department of Public Health, Yamagata University, Japan. He holds a concurrent commitment as a consultant ophthalmologist, specializing in diagnosis and treatment of retina and macular conditions, including diabetic retinopathy, age-related macular degeneration, retinal vein occlusion, pathologic myopia and other vitreo-retinal diseases.
Dr. Kawasaki has been involved in wide range of ophthalmic epidemiology research, including preventive medicine, standardization of clinical diagnostic schemes of retinal diseases, utilizing quantitative retinal measurements as a marker to better predict development of eye diseases and systemic diseases.
He is a Principal Investigator of multi-centre trials and population-based cohort studies supported by NHMRC and Japan Society for the Promotion of Science Project grant. Dr. Kawasaki is on the Editorial Board of Ophthalmic Epidemiology.
His current passion is to establish a national screening program for diabetic retinopathy integrating clinical epidemiological evidences and quantitative retinal measurements aiming to preserve better vision, beyond preventing blindness.
Host: Professor Emanuele Trucco - 01/05/2013: Mr. Rajnikant Sharma - Enhancing Business Efficiency Through ICT
KITE Innovation is participating in a Knowledge Transfer Partnership
(KTP) with the University of Dundee. This seminar will provide a brief
overview of the different ICT tools that were identified in this KTP as
being of key importance for the KITE business. New ICT tools have since
been developed to support the business processes of KITE.
Comparison will be made with the way the company used to perform their
work before the advent of ICT Systems. How the business of the company
is affected by Integration of new ICT tools will be described. The
future plans of the project will also be discussed.
Bio:
Rajnikant Sharma has been employed as a KTP Research Associate with the
School of Computing, Dundee University and KITE Innovation Europe LTD
since February 2012 on a 27-month university/industry project. His role
includes all aspects of requirements capture, design, software
development and evaluation. He is also responsible for customer liaison
and project reporting.
Rajnikant has a BSc (Hons) in Computing from Teesside University in
2011. His Honours thesis was entitled "FoodStop - An e-commerce website
for Online food ordering using ASP.net and MS-SQL". - 18/04/2013: Prof. Bin Luo - Image Modeling Using Geometric-Edge Random Graphs
In this talk, I will concentrate on random graph-based methods for digital image processing and pattern recognition. A random graph model for image representation, which is recently developed in our group, will be introduced. The model is noted as Geometric-Edge (G-E) Random Graph Model. Then we cast image matching into G-E Random Graph matching by using the random dot product graph based matching algorithm. Experimental results show that the proposed G-E Random Graph model and matching algorithm are effective and robust to structural variations.
BIO:
BIN LUO is present the dean and a professor of the School of Computer Science and Technology, Anhui University, China. He received his BEng degree in electronics and MEng degree in computer science from Anhui university of China in 1984 and 1991, respectively. In 2002, he was awarded the Ph.D. degree in Computer Science from the University of York, the United Kingdom. From 1996 to 1997, he worked as a British Council visiting scholar at the University of York under the Sino-British Friendship Scholarship Scheme (SBFSS). He joined the University of York as a research associate from 2000 to 2004. He was a short time research fellow of British Telecom in 2006. In 2007, he visited the University of Greenwich, England as a visiting professor. He served as a visiting fellow of the University of New South Wales, Australia in 2008. In 2009, he was a visiting professor at the Florida Institute of Technology. He held a position of TCT Exchange Fellowship at Nanyang Technological University, Singapore in 2012. He is present a visiting professor at the University of Stirling, UK. He has published more than 200 research papers in journals, edited books and refereed conferences. Some of them were published on IEEE T-PAMI, Pattern Recognition, CVPR, IJCAI, AAAI. His research interests include graph models for image representation, image and graph matching, graph spectral analysis. He chairs the IEEE Hefei Subsection and is a senior member of IEEE.
Acknowledgement: This talk is supported by school of computing
Host: Dr. Jianguo Zhang - 27/03/2013: Miss. Heather Ellis - A discussion on my research for eHealth systems to support caregivers of complex epilepsy
As eHealth techonologies become more dominant in today's technology driven society, there becomes more products which better support the care of those with long term conditions. However, those with complex epilepsy are not supported by the current systems. Those with complex epilepsy are more likely to need more support from caregivers (both informal and formal) during their lives. I will be discussing how I led a design study last summer that looked at the user interface of such a system. The session will discuss the results of the study and my future goals for this project .
Bio:
Heather Ellis is currently studying BSc Applied Computing at the University of Dundee and is in her third year. She has recently published an Student Research Competition paper at CHI about and her current research interests include Human Computer Interaction and how we can develop new eHealth systems to better support the informal caregivers of those with longterm health conditions. Homepage: heatherellis.net.
- 13/03/2013: Dr. Ardhendu Behera - Real-time Activity Monitoring and Recovery using Wearable Sensors
The research involves a novel approach for real-time activity monitoring and recovery in an egocentric setup using on-body sensors of IMUs (Inertia Measurement Unit) and camera. In a given activity, the component atomic events are characterised in terms of spatiotemporal pairwise relationships between body parts, objects, tools and machine parts in a workspace. The motivation is that these relationships are invariant to position and viewpoint, and correlate well with functional relationships between objects (e.g. picking up an object with the hand involves contact between the body part and the object; hitting a nail with a hammer involves a high speed of approach between the hammer and the nail).
We have combined the above-mentioned representation for the key objects in the workspace with an HMM to model workflow activity and observation uncertainty. Details of this model and associated evaluation can be found in Behera, Hogg and Cohn in ACCV 2012. A workflow activity is assumed to be a temporally ordered set of procedural steps or atomic events for accomplishing a task. The atomic events are associated with the states of an HMM, with a conditional observation distribution over a summary of the expected spatiotemporal relationships between pairs of objects in the workspace within a sliding time window. These objects include parts of the human body such as the wrists, work tools such as hammers and screwdrivers, and machine parts. The observation distribution is represented by a probabilistic multi-class SVM. The key novelty in our model is the way in which we characterise the movement and interaction of a varying number of objects using a fixed observation space, and in such a way that recognition performance is invariant to broken tracks and missed detections.
Bio:
Ardhendu Behera has received his PhD degree in Computer Science at University of Fribourg, Switzerland. Soon after, he has joined as Research Fellow on the EPSRC HAML (Cognitive Foresight: Human Attention and Machine Learning) research project in the Computer Vision group, at the University of Leeds UK. Currently, he is working as a Research Fellow on the EU FP7 COGNITO (Cognitive Workflow Capturing and Rendering with On-body Sensor Networks) research project in the same group. His research focuses on Activity Analysis and Recognition, Scene Understanding and Visual Attention. Prior to his PhD, he has worked in Sun Microsystems as a member of technical staff. For more information, please refer to http://www.comp.leeds.ac.uk/behera/
Host: Professor Stephen McKenna - 04/03/2013: Professor. Henning Christiansen - The Experience Cylinder at Roskilde and a first step towards rule-based a programming of knowledge i
The Experience Cylinder is a generic, interactive installation currently under development at Roskilde University. It consists of a large circular screen, a number of synchronized projectors, a Kinect device that tracks the users and a loudspeaker system for creating localized sound. It was first developed for an application in cooperation with the Viking Ship Museum of Roskilde for visualizing data collected during a voyage from Roskilde to Dublin and back with a full size reconstruction of a Viking longship.
The software is now being developed into a more generic form, where new active components can be added easily, almost in a drag-and-drop fashion. In relation to this, we are experimenting with an adaptation of the language of Constraint Handling Rules for programming of knowledge intensive components in such installations. The goal is to provide simple and rule-based programming facilities for distributed reasoning with persistent and shared knowledge bases, through which communication between different components also can be made.
The speaker will also give an overview of other selected projects in his research group PLIS: Programming, Logic and Intelligent Systems, a research group at Roskilde.
- 27/02/2013: Dr. Ekaterina Komendantskaya - A talk about SICSA grants for PhD students and postdots
The Scottish Informatics and Computer Science Alliance (SICSA) is a collaboration of Scottish Universities whose goal is to develop and extend Scotland's position as a world leader in Informatics and Computer Science research and education. SICSA has a renewed funding directed to facilitating early career researchers' exchanges with business and industry, both within Scotland and internationally. SICSA Early Career Researcher (ECR) Board helps to guide on the allocation of restored pool funding worth £410,000 in 2013-2015. I am a SoC's representative at the SICSA ECR Board, and in this brief (30 mins) talk, I will tell about various schemes available from SICSA. See also http://www.sicsa.ac.uk/funding.
Bio:
Katya Komendantskaya is a Lecturer at the School of Computing, for more details see http://www.computing.dundee.ac.uk/staff/katya/
- 08/02/2013: Dr. Weishi Zheng - Tutorial and Discussion on Convex Optimization and Robust Learning in Pattern Recognition
Abstract:
In the past decade, half-quadratic (HQ) optimization has become increasingly popular for solving computational problems in sparsity estimation and robust learning, which is important for computer vision, image processing, and pattern recognition. In this tutorial and discussion, I will some basic theory and techniques of HQ optimization, as well as its applications in pattern recognition.
This tutorial is part of the course that gave in the ICPR 2012 conference: http://www.icpr2012.org/tutorials-AM-02.html
The brief agenda will be:
1 - Introduction and Overview of the Tutorial (10 minutes)
2 HQ Optimization in Robust Learning (45 minutes)
2.1 HQ Optimization in Image Denoising
2.2 HQ Optimization in Subspace Learning
2.3 HQ Optimization in Feature Extraction
Discussion
3 HQ Optimization in Sparsity Estimation (35 minutes)
3.1 A Brief Review of HQ Optimization
3.2 L1 Minimization and Robust Sparse Representation
3.3 HQ framework for Robust Sparse Representation
Discussion
Bio:
Wei-Shi Zheng received his Ph.D. degree in Applied Mathematics at Sun Yat-Sen University, China, 2008. After that, he has been a Postdoctoral Researcher on the European SAMURAI Research Project at the Department of Computer Science, Queen Mary University of London, UK. He is now an associate professor of Sun Yat-sen University, and a selected member of the One-hundred Talents Program of Sun Yat-sen University. He has published widely in top journals and conferences including IEEE TPAMI, IEEE TNN, IEEE TIP, Pattern Recognition, IEEE TSMC-B, Neural Computation, IEEE TKDE, Neural Networks, ICCV, CVPR and AAAI. His current research interests are in object association and categorization for visual surveillance and large scale machine learning problems. He gave tutorial on the ICPR 2012 conference. Homepage: http://sist.sysu.edu.cn/~zhwshi/
Acknowledgement: This talk is supported by school of computing and The Royal Society of Edinburgh (RSE-NSFC) joint project.
Host: Dr. Jianguo Zhang - 06/02/2013: Dr.Mike Massimi - Designing and Developing an Online Support Group for the Bereaved
Designing interactive systems that sensitively engage with end of life issues is a key challenge for human-computer interaction as more and more people turn to the Internet to grapple with the realities of death. However, there are few studies that document how specific remembrance or support features of bereavement websites are used in a real-world setting. In this talk, I’ll describe Besupp, a website where bereaved individuals can participate in online support groups. Three support groups used Besupp in a ten-week long deployment study. Based on this study, I describe how participants perceived and used the remembrance and social support features of the site. These results form the basis for a set of implications regarding the design of technologies for remembrance, social support, and bereavement more generally.
Bio:
Michael Massimi is a Post-Doctoral Researcher in the Socio-Digital Systems group at Microsoft Research in Cambridge, UK. His research focuses on technology’s role during life-changing events such as marriage, parenthood, and the death of a loved one. His work has explored the role of technologies for social support, communication, and memory as they relate to these transitory times. He is an alum of the Health Care, Technology, and Place doctoral training program at the University of Toronto, where he earned a Ph.D. in Computer Science. For more information, please see http://research.microsoft.com/en-us/people/mmassimi/
Host: Dr. Wendy Moncur - 04/02/2013: Dr. Wei-Shi Zheng - Distance Learning in Person Re-identification
In this talk, I would introduce distance learning techniques and discuss their use in person re-identification. Person re-identification is to tell whether two person images captured in non-overlapping camera views are from the same person. Due to large appearance change, view change, lighting variation, severe occlusion and etc., it is a big challenge to compute the similarity between images. We evaluate existing popular distance learning methods and propose relative distance comparison models. Our investigation suggests the relative comparison model is indeed a discriminative and robust way for similarity measurement in person re-identification.
Bio:
Wei-Shi Zheng received his Ph.D. degree in Applied Mathematics at Sun Yat-Sen University, China, 2008. After that, he has been a Postdoctoral Researcher on the European SAMURAI Research Project at the Department of Computer Science, Queen Mary University of London, UK. He is now an associate professor of Sun Yat-sen University, and a selected member of the One-hundred Talents Program of Sun Yat-sen University. He has published widely in top journals and conferences including IEEE TPAMI, IEEE TNN, IEEE TIP, Pattern Recognition, IEEE TSMC-B, Neural Computation, IEEE TKDE, Neural Networks, ICCV, CVPR and AAAI. His current research interests are in object association and categorization for visual surveillance and large scale machine learning problems. Homepage: http://sist.sysu.edu.cn/~zhwshi/
Acknowledgement: This talk is supported by school of computing and The Royal Society of Edinburgh (RSE-NSFC) joint project.
Host: Dr. Jianguo Zhang - 29/01/2013: Michael Marra - Design in Action: Knowledge Exchange and the Scottish Economy
Design in Action is a unique opportunity to turn academic research into an economic driver for Scotland. This seminar will introduce the Design in Action model that is being led by DJCAD here at the University of Dundee and the opportunities that it presents for academics and for Scotland at large.
As Scotland seeks elusive paths back to economic growth a dynamic model of knowledge exchange offers one important possibility. By re-positioning design from an aesthetic to a strategic practice Design in Action seeks to generate radical innovation that is near to market and can create rapid economic value. Through the development of the new 'Chiasma' methodology, DiA's radical innovation residential, interdisciplinary teams are brought together to answer complex real world problems. The first Chiasma will be held in February 2013. Applications to attend can be made at http://designinaction.com/
The seminar will afford colleagues the opportunity to discuss the methodology, its varied applications and the prospects for involvement and new research outputs.
BIO:
Michael Marra MA Msc, Deputy Director, Design in Action
Michael studied History at undergraduate level at the University of Glasgow and trained as a development economist at the London School of Economics and Political Science. He was head of Policy and Public Affairs for Oxfam Scotland before taking up a post as Senior Political Adviser to the Leader of the Opposition in the Scottish Parliament. For two years he led on political and policy matters across all portfolios with particular interests in economic development, energy and poverty issues. He took up the post of Deputy Director in March 2012 and has assembled a team of 26 staff across six partner universities. DiA is now the largest design research unit in the UK.
- 18/01/2013: David Price - Visualising Public Policy
This presentation, given by on of the founders of DebateGraph, illustrates some of the ways in which DebateGraph is being used in a public policy context, and reflects on the insights arising from this experience.
Host: Prof. Chris Reed - 16/01/2013: Dr Alex Dunedin
Title of talk
The Fantasy Enlightenment League; Using available digital infrastructure to create an open peer review system
Content of talk
Ragged University was set up to develop inclusive forms of bonding, bridging and linking social capital (trusting networks) using free education as a lens to focus community. As a voluntary project we work in the third sector, and develop close working relationships with academia and industry. The Ragged website acts as a virtual platform for promoting events and activities which are free to attend and open to all, publishing peoples work and highlighting free knowledge resources.
The Fantasy Enlightenment League is a scheme devised to make everyone a stakeholder in knowledge in a fun way, fostering invention and innovation through a dynamic learning environment.
Key priorities are:
*
Incentivising knowledge production in the community (artifacts)
*
Creating learning frameworks which provide pathways through recapitulation (histories)
*
Generate situations of open peer review (collaborative ecologies)
*
Use of common technologies to engage a wider population (publishing media)
There is a great need for people to grapple with the rathionales involved in how we arrive at ideas and increasingly reliable knowledge. There is a necessity to have the fact that everyone is a part of intellectual life of first principles and analytical thought as common currency. It is important culturally that the message of ‘we are all capable of learning and contributing to collective conversations’ exists in a tangible form. This has extended effects in the social, economic and educational realms.
Over time the question of how the Ragged project gets people to engage in knowledge production (i.e. theories) and communitive activities (i.e. peer review) whilst learning key concepts in knowledge building has been explored. This has been discussed with a number of professionals, communities, and academics. This is a discussion of the Fantasy Enlightenment League and the technologies being explored to realise this vision
Bio
Alex Dunedin has a background as a library researcher developing work committed to the public domain, primarily in the bio-medical sciences area. When Alex Dunedin heard about ‘ragged’ schools, which brought free education to the United Kingdom in the 19th century, he decided to update the movement using available infrastructure and common technology. Ragged University provides free talks and debates in informal spaces such as pubs, cafes and libraries, featuring speakers from all backgrounds and ages. It aims to complement, rather than replace, schools and universities, many of which have already pledged their support to the project. Alex is self-educated and places autodidactism at the forefront of his work alongside personalised good working relationships.
Host: Prof Emanuele Trucco - 19/12/2012: Dr. Atsuto Maki - Co-occurrence Flow for Pedestrian Detection
Title: Co-occurrence Flow for Pedestrian Detection
The talk first introduces a new type of motion feature, the
co-occurrence flow (CoF), which we designed for pedestrian detection.
While recent work has established a combination of oriented gradients
and optic flow as effective features, our advance is to capture
relative movements of different parts of the entire body, unlike
existing motion features which extract internal motion in a local
fashion. Through evaluations on the TUD-Brussels pedestrian dataset,
we show that our motion feature based on co-occurrence flow
contributes to boost the performance of existing methods.
During the talk I will also brief some other recent research
activities on computer vision, taking place at Toshiba Research Europe
- this will include:
(a) a new algorithm to generate homogeneous superpixels from the
process of random walks. We exploit the Markov cluster algorithm as
the methodology and, in particular, a new graph pruning strategy
called compact pruning in order to capture intrinsic local image
structure. The superpixels are thereby kept homogeneous, i.e. uniform
in size and compact in shape; an optimal performance is achieved in
terms of qualitative measure at a decent computational speed. Joint
work with F. Perbet.
(b) a new similarity measure, the sum of conditional variance of
differences (SCVD), designed to be insensitive to highly non-linear
intensity transformations such as the ones occurring in multi-modal
image registration and tracking. It improves on another recently
introduced statistical measure, the sum of conditional variances,
which has been reported to outperform comparable information theoretic
similarity measures. Joint work with Riccardo Gherardi.
BIO
Atsuto Maki received his PhD degree in computing science from the Royal Institute of Technology, Stockholm, Sweden, in 1996. Prior to that he studied electrical engineering at Kyoto University and the University of Tokyo, Japan. He is currently a Senior Researcher at Toshiba Research Europe, Cambridge, UK. His recent research interests cover a broad range of topics in machine learning and computer vision, including motion and object recognition, clustering, registration, and stereopsis. He was an Associate Professor at the Graduate School of Informatics, Kyoto University, until 2008. He has been serving as a program committee member at major computer vision conferences including an Area Chair of ICCV'09.
Host: Prof Emanuele Trucco
Any questions, comments or speaker suggestions? Contact Hu Xu.