Events

TBA

TBA

In dynamic graph drawing, preserving the mental map, or ensuring that the location of nodes do not change significantly as the information evolves over time is considered an important property by algorithm designers. Many prior experiments have attempted to verify this principle, with surprisingly little success. These experiments have used several different algorithmic methods, a variety of graph interpretation questions on both real and fabricated data, and different presentation methods. However, none of the results have conclusively demonstrated the importance of mental map preservation on task performance. Our experiment measures the efficacy of the dynamic graph drawing in a different manner: we look at how memorable the evolving graph is, rather than how easy it is to interpret.

As observed in the previous studies, we found no significant difference in terms of response time or error rate when preserving the mental map. While preserving the mental map is a good idea in principle, we find that it may not always support performance. However, our qualitative data suggests that, in terms of the user’s perception, preserving the mental map makes memorability tasks easier. Our qualitative data also suggests that there may be two features of the dynamic graph drawing that may assist in their memorability: interesting subgraphs that remain visible over time and interesting patterns in node movement. The former is supported by preserving the mental map while the latter is not.

In April I went to ECIR 2012 for my poster and worked at WWW 2012 as a student volunteer. I will report back my experiences there.

With the advancement of data storage technologies and the ever-decreasing costs of hardware, our ability to store data is unprecedented. However, a large gap remains between our ability to collect and store data and our ability to derive knowledge from it. Data visualization exploits our most powerful sense, vision, in order to derive knowledge and gain insight into large, multi-variate data sets that describe complicated and often time-dependent events. This talk provides a brief introduction to data visualization, its strengths, weaknesses, and, goals. We also describe a selection of inter-disciplinary projects where visualization is used to address fundamental questions-the answers of which we hope to discover in various large, complex, and time-dependent phenomena.

TBA

TBA

TBA

Symmetries are prevalent in nature and man-made objects. In this talk I will discuss rotational symmetries, an important class of symmetries, and their applications in computer graphics such as non-photorealistic rendering, texture and geometry synthesis, street modeling, and quadrangular and triangular remeshing. I will also review important mathematical analysis related to rotational symmetries on surfaces as well as discuss possible future directions to 3D.

When experienced programmers solve non-obvious performance problems in Haskell, their first stop is often a look at the "Core" output of the compiler. This intermediate transformation stage has proven to be quite illuminating when trying to understand a program's performance properties. However, due to its disconnect with the source code, often daunting complexity as well as non-obvious performance implications, only experts have so far been able to perform this kind of analysis.

Our goal is to change this by actually putting Core front and centre in a program optimization tool. Our program would help the user navigate a representation of the Core code as well as associate it with all relevant source as well as performance data. This should put the user into a position to make informed decisions about how to change the program in order to obtain a translation with better performance properties.

This talk is about a work in progress, so we will unfortunately not be able to show any finished results. We would however appreciate a discussion of the approach and would welcome ideas on tackling the visualization issues.

In the previous talk, I presented results of my empirical user study to investigate revisiting patterns, potential useful cues, and underlying causes of failure when people revisit non-frequently and non-recently visited webpages. Based on those findings, a new visualisation tool has been designed to address main underlying causes of failure. In this talk, I will be describing the design and bring it up for discussion.

TBA

Andrew Davies, Matty Wall, Ryan Cundell.

John Lau, Alexander Jones, Andrew Rae.

Tom Lumbard, Paul Malone, Haroon Sheikh.

TBA

TBA

Over the last two and a half years I have been working on a project the aim of which is to create a Virtual Reality Microscope that is suitable as a replacement for a conventional light microscope. Though digital slides have several advantages over traditional light microscopes, diagnoses using digital slides have been shown to be up to 60% slower that diagnoses using traditional microscopes. In this talk I will provide an overview of my part in this project.

My presentation will be to introduce myself and my work to you, more importantly to get your views on my research.

Geology students need strong spatial awareness and skills to be able to visualize the geology they study. Despite the potential of current tablets and smart-phones in terms of hardware and software, there has been little research into the use of such devices for educational geology fieldwork.

During my research I will try and find answers for the above problem within the new tablet/mobile world of computing for visualization issues which are rooted deeply within student geologists and the wider professional geologists as well.

Although huge amounts of geologic data is available, desktop computers have been and are being used to visualize such data for decades. "Capturing data using devices" and "taking data back to the field" are the starters of visualizing such data in the field or at least taking the "visualized" data on desktop computers onto mobile/tablet devices back into the field.

My research question is: How can tablet/mobile devices take the step taken by desktop computers further afield?

While investigating a number of successful meta-heuristics for the Traveling Salesperson Problem, a number of patterns of computation were noticed. These patterns give rise to a variety of subtly varying search strategies, and it was resolved to see how their performance varied. Using the Haskell framework for meta-heuristics, which has been the key focus of my PhD, a short program was created and left running, combining these small building blocks, testing them and recording the results. In this talk I will be showing the pattern of the results that emerged and how it inspired the modification of a meta-heuristic originally designed for a large real world problem, with promising results.

Most people path-find through familiar urban environments every day with very little conscious thought or effort, but what happens if this activity is disrupted? What if they encounter an unfamiliar area or if there is a deterioration in the cognitive abilities underpinning this process? In many cases maps, whether paper or those widely available on the internet, are used to plan routes through unknown cities or towns. Unfortunately, these representations may not contain sufficient information for pedestrian travel, or the individual concerned may not be capable of translating it successfully into a route. The proposed research aims to create a system which not only provides routes in these situations, but also encourages the user to learn some of the spatial and location information required to form a cognitive map of the environment.

This initial presentation will give an overview of the direction my PhD will take, the concepts that will be important and the work completed so far. It will consider the navigational tasks encountered by a range of individuals and how, by approximating human strategies and approaches, suitable routes can be generated for some of these scenarios.

TBA

TBA

TBA

TBA

TBA

A fundamental problem in network visualization is limited display space. Basic pan and zoom tools enable us to focus on parts of the network but visual references into the surrounding context are lost. To move to a new region of interest the user must disengage from analysis and perform a series of navigation tasks: zoom out and pan to gain context; locate the new region of interest; zoom in and pan to focus on detail. Only then can the primary task of data analysis resume. Aside from transforming the data by filtering and clustering, solutions to this problem have largely concentrated on providing multi-scale views of the network, either in seperate windows or by non-uniform scaling of the display space.

An alternative approach is "off-screen visualization", that is, on-screen information about the location of off-screen regions of interest. In this talk I identify the key issues of off-screen visualization design for networks and present a taxonomy of the design space based on the existing literature. This classification situates my current work (see CoronaScope) and also serves to identify possible future directions.

Hannes will report back from VisWeek 2011, Providence, RI.

TBA

TBA

TBA

TBA

Scientific computation is most commonly used to simulate continuous phenomena. Scientific visualization, therefore, depends very heavily on continuity in the mathematical models to be applied. Thus, not only do good visualizations depend on good data, they depend on understanding and reproducing the assumptions about continuity that underlie the computation in the first place. Moreover, some implications of continuity that are not part of conventional numerical analysis become increasingly important when we attempt to present large-scale data to a human viewer.

One of the stages where this occurs is presenting a discrete representation based on particles, voxels or meshes to the human eye. Since the human eye expects visual continuity, it becomes important how continuity is reintroduced after the computation. Another stage where this occurs is in topological analysis - as data gets ever more massive, more sophisticated forms of analysis are invoked, and these too depend on assumptions about continuity.

The presentation will therefore cover the fundamental role of continuity in how we approach scientific visualization, some examples of where the choice of models of continuity causes real difficulties, and a discussion of some problems where the assumptions of continuity in visualization and analysis are in tension with the assumptions of continuity in numerical simulation.

Details to be provided.

The task of program optimisation is frustratingly hard, with many possible problem sources to consider. Consequently, good profiling tools are essential for observing program behaviour and helping the programmer to spot the patterns and underlying reasons.

The ThreadScope profiling utility has already proved invaluable in this task. However, so far it has mainly tracked performance characteristics, with little help for relating them back to the program in question. Our aim is to improve this situation by allowing the user to relate the observed behaviour to locations in the application's source code.

We have extended the GHC tool chain accordingly and are now able to give quick and accurate program profiles, complete with source code references. Our talk will give an overview of the new functionality and its implementation. We will talk about:

• The profiling back-end, which allows sampling code pointers according to different performance characteristics,
• How we manage to retain code associations throughout the various code generation stages of GHC,
• And the user interface we have designed to make the generated profiling data accessible to the user.

TBA

TBA

My PhD addresses the problem of how to use a visualization tool to help people revisit webpages that have been visited neither frequently nor recently. To design an effective tool, a user study has been done to investigate how people revisit webpages, what difficulties they might encounter, and what information might be useful for them. In the first part of the talk I will describe the use study and report its main findings. Then some pilot design will be presented for discussion.

TBA

Where performance is the primary objective, it is still rare to find functional languages being used. While improved compilers have lead to a number of remarkable success stories, real-world usage suggests that taking the imperative approach is still more likely to lead to a fast program right away. This talk will try to dig out some of the reasons for that, before investigating how the situation might well reverse in future.

GP is a rule-based, nondeterministic programming language for high-level problem solving in the domain of graphs, freeing programmers from handling low-level data structures. The language has a simple syntax and semantics, to facilitate formal reasoning about programs. In this talk, I will introduce GP by a number of example programs and discuss its structural operational semantics. A special feature of this semantics is the use of failing computations to define branching and iteration. I will then present proof rules for Hoare-style verification of graph programs, and demonstrate their use by verifying a simple colouring program. Finally, I will briefly describe GP's current implementation.

TBA

TBA

TBA

TBA

Increasingly in industry, manufacturing considerations are being made early in the product design process. Manufacturing engineers and Design engineers collaborate using digital tools to prototype the manufacturing and assembly process and identify potential product design changes to aid in cost effective manufacturing. Virtual reality has a key role to play in providing a common communication medium for idea exchange between these groups of experts. This presentation will explore the challenges of using virtual reality for manufacturing and assembly prototyping and design. New approaches which include using voxel-based collision detection and b-rep-based geometric constraint enforcement along with 6 degree-of-freedom haptics look promising as a means to support virtual assembly of low clearance CAD models. Current research performed at Iowa State University in the Virtual Reality Applications Laboratory will be described along with visions for future research.

In my previous work (presented here on 3rd March 2011 and now known as "CoronaScope" I demonstrated a system for navigating large networks (graph drawings) based on a landmark overlay and a central widget that indicates the distance and direction to off-screen landmarks. The landmarks were defined a priori and I offered only limited guidance on how the landmarks could be selected, leaving the problem for the domain expert.

Recently I implemented a method of selecting landmarks interactively during exploration of a graph. In this talk I will briefly review CoronaScope before demonstrating the new functionality.

Joint work with David Duke.

One form of topological analysis of multi-variate data extends Reeb Graph / Contour Tree analysis through Jacobi sets, which study the systematic variation of multiple variables. Algorithms for computing Jacobi sets are difficult to define, in part because isovalues chosen for each variable reduce the dimensionality of the resulting sets. This can be stepped around by quantizing the domain (as is implicit in computation anyway), then studying the graph relationships of the partial manifolds of full dimension.

TBA

TBA

TBA

Rebecca will give a talk on designing and conducting user studies for the design of visualisation systems.

By: Zhongzang Lin, Harry Yeh, Robert S. Laramee & Eugene Zhang

In this article we define the topology of 2D asymmetric tensor fields in terms of two graphs corresponding to the eigenvalue and eigenvector analysis for the tensor fields, respectively. Asymmetric tensor field topology can not only yield a concise representation of the field, but also provide a framework for spatial-temporal tracking of field features. Furthermore, inherent topological constraints in asymmetric tensor fields can be identified unambiguously through these graphs. We also describe efficient algorithms to compute the topology of a given 2D asymmetric tensor field. We demonstrate the utility of our graph representations for asymmetric tensor field topology with fluid simulation data sets.

In this presentation I will share some of my experiences of building Powerwalls at St. James' Hospital, Leeds. I will discuss the hardware used and some of the problams that arose during construction. I will then present some of the ways in which the performance of the Powerwalls has been evaluated using data derived from physical microscope use.

TBA

Creating a Visual Interface for File System, David Richardson: The widely popular desktop metaphor has been around for 30 years and since then a lot has changed. While interfaces such as Windows Explorer have evolved over the years, perhaps a completely new interface is needed to meet the demands of today's users.

Force-Directed Graph Layout with Parallel Haskell, Rob Marshall: Force-directed layout algorithms aim to produce aesthetically pleasing visual representations of graphs using formulae inspired by physics. In this presentation, I aim to discuss popular algorithms, present a demo of a simple Haskell implementation, and discuss strategies to increase performance by running on multiple cores.

In previous work we demonstrated the use of explicitly defined landmarks to provide navigational reference points in large, complex graph drawings [1]. One of the problems with that work was that landmarks are lost off-screen when the user uses zoom and pan to explore the graph. To counter this we introduce a new technique for off-screen visualization by providing an on-screen representation of landmarks while aiming to avoid visual clutter. We are also exploring the use of automated pan and zoom -- watch this space!

I will present the theoretical basis for our design, and demonstrate our prototype software. As a further incentive I will provide a range of biscuits.

[1] Colin Myers and David Duke, A Map of the Heap: Revealing Design Abstractions in Runtime Structures. In Proceedings 5th ACM International Symposium on Software Visualization (SOFTVIS), 2010.

Visualizing Parameter Space, Dominic Gray: Finding an optimal set of parameters from a given range can be a difficult task when dealing with a high dimensional parameter space, so finding a way to visualize the parameter space in a way that highlights optimal regions can be very beneficial. I plan to give a brief summary of what my project involves, and talk about some of the current methods of parameter space visualization being used. I will then present my ideas so far; looking at paramater optimization in the manufacturing field to optimize/maximize profit, and I will show some functions I am using to plot profit curves for different parameters, and how they link together to represent overall profit.

Can functional programming's parallel karma tame computational topology's fearsome dogma? Problems in computational topology, such as generating the contour tree for a dataset, have proved to be stubbornly sequential. Some parallel solutions are known, but the methods are complicated. Recent progress in parallel functional programming *may* offer alternative lines of attack, for example using fine(r) grain parallel evaluation of sub-expressions. In this talk I will report on some initial work exploring the implementation of Hamish Carr's contour tree algorithm in Haskell.

The results are interesting: while Haskell can match the speed of a simple imperative sequential implementation, straightforward use of fine-grain parallel operators produces limited scalability. I will discuss future directions for the work, and some general lessons about high-performance parallel Haskell.

Collaboration between users relies on them having a common grounding of knowledge to understand other users' actions. When this breaks down the collaboration cannot continue until it is repaired. These breakdowns can increase as elements such as the display used to view the data differ between users. This presentation will highlight the effect the use of a Powerwall display by one collaborator has on common ground breakdowns in Collaborative Virtual Environments (CVE's) before discussing techniques to reduce these breakdowns for desktop users.

Discussion of crossover research ideas between the VVR and NLP research groups.

TBA

TBA

TBA

TBA

TBA

TBA

TBA

Colin will share some of his recent experiences at VisWeek 2010, held from 24-29 October in Salt Lake City, Utah. Hamish Carr will also share some thoughts on the event.

I'll give a short overview of the problem as we currently see it with reference to CellProfiler, an image analysis tool developed by the Broad Institute. I'll show initial work on sampling input parameters and generating resulting output with CellProfiler and go on to talk about a prototype I'm currently working on for browsing and analysing this output.

An attempt to generate 2d visualisations of local search graphs (7 and 8 connected), using the TSP as a test problem. A presentation of some of the visualisations that have been created, and a discussion of why they do not help in any way at all.

Hosted by Hamish Jamson

This impetus for this talk is Roy and my work on getting to grips with image analysis parameter space, which is multidimensional. I will present an overview of established techniques, but the aim is also to spark informal discussion.

Digital microscopy is a technology that has the potential to revolutionise work practice within histopathology, with promised improvements in safety and efficiency. Despite this, uptake of the technology is slow and previous studies report scepticism and uncertainty amongst potential users. In order to design a successful digital microscope - one which fits with the work practices of histopathologists, which they are happy to use within their daily work, and which leads to an improvement in the processes of care - an understanding of their current work practices and the context within which they carry out their work is required. It is important to understand the work of diagnosis within its real-world context, as opposed to how the work of diagnosis takes place within an experimental setting. It is also important to understand the work of diagnosis in its entirety; diagnostic work does not necessarily begin and end at the microscope.

As part of a project to develop a digital microscope, we conducted naturalistic observations that explicate the detail of how histopathologists accomplish their work, in order to produce an account of work practices in histopathology that extends beyond the microscope. Observations were combined with semi-structured interviews, to capture histopathologists' perspectives on their work and their attitudes toward digital microscopy. I will present the findings of our studies and consider the implications of this for the design of a digital microscope.

As part of the UKVAC (United Kingdom Visual Analytics Consortium), Middlesex University held VASS (Visual Analytics Summer School) from 16-24 September 2010 at their Hendon campus in London. This residential event was sponsored by the US Department of Homeland Security via the US National Visualization and Analytics Centre (NVAC) and was organised in collaboration with the UK Home Office and Ministry of Defence. Over a seven day programme delegates learnt about visual analytic problems, techniques and solutions from leading international experts in the field. Included in the syllabus were: analytical reasoning; the design of interactive visualisations; data representations and transformations; perspectives on software engineering and programming for rapid visualisation. The aim of the summer school was to provide a multidisciplinary grounding on this emerging and important area. Trien will report back on his experiences at VASS 2010.

Visual representations of runtime software structures such as heap memory graphs can aid in debugging and help to develop program understanding. However, such structures may contain thousands of objects and have no obvious spatial organisation. If the program contains flaws the appearance of objects may well differ from the user's expectations. Navigating these graphs can be challenging to the user as the space is abstract and potentially unfamiliar.

To alleviate this problem we employ a systematic approach grounded in the principles of navigational landmarks. We identify subgraphs within the heap that correspond to significant design abstractions and apply various visualization techniques to highlight and organise these structures. The aim is to provide the user with recognisable features that are linked to more familiar representations of the software. We claim that the enhanced representation can support existing memory debugging tools by providing the user with a usable 'map' of an otherwise abstract data space.

The results are demonstrated using data extracted from an instrumented version of the Visualization Tool Kit (VTK), a complex and widely-used architecture for data visualization.

This talk is my first run in preparation for SoftVis and a copy of the accompanying paper can be found here. I am also presenting a poster at InfoVis that describes some of the theoretical basis for my work in greater detail. A copy of the poster summary can be found here.

I will review how "history" mechanisms have been used in 25+ years of visualization applications, and then focus on the recently developed VisTrails system.

Image analysis software, like many computer-based systems, often have a large number of input parameters that influence their output. Tweaking parameters to find appropriate settings is an acknowledged usability problem. Interactive visualization has been shown to be a useful approach for addressing this challenge. In this talk, I will outline the problem, present existing research results and introduce our own preliminary work on parameter visualization.

This talk will outline some *very* preliminary ideas about interaction between machine learning (ML) and visualization. ML algorithms such as k-means clustering and self-organizing (Kohonen) maps are already used in some vis applications, and further interest has been prompted both by the visual analytics agenda and the need to cope with data scale. My particular interest is in a more exotic sub-field of ML, inductive logic programming (ILP). I want to suggest that ILP could play an important role in information visualization, providing a clearer theoretical framework for understanding and evaluating infovis, and new practical tools for addressing the problems of scale.

Isosurface extraction is a widely exploited visualization technique for volumetric data on all manner of grid representation. The basic technique is often used to explore and measure many properties of data sets of ever increasing size. Therefore, data structures and algorithms that facilitate interactive exploration and fast processing of isosurfaces of large data sets is of paramount importance. While many optimal methods have been proposed to accelerate isosurface extraction, many of these algorithms have limitations with regards to storage costs and data quantization. In some cases these limitations preclude their practical application. We present a very simple clustering and volume compression technique based on observations in the span space and show that applying this technique to existing methods can reduce their storage cost. We show results for real data validating our technique.

TBA

TBA

TBA

Roy will share some of his experiences at IV 10 in London.

Medial axes, or skeletons, are well-known shape descriptors. For decades, they have been developed and used in the computer vision community for applications such as shape retrieval, matching, simplification, and recognition, and path planning. Recent advances indicate that medial axes have great potential beyond computer vision, e.g. in 3D scientific and medical visualization and image and data compression. However, several theoretical and practical aspects need to be addressed.

In this talk, we focus on recent developments in 3D multiscale medial axes. We present several examples of these recent developments in multiscale medial representation, robust computation, and applications in geometry processing and data visualization. Secondly, we give examples of potential cross-disciplinary applications where medial axes can bring added value and open new research directions, such as multiscale and multivariate signal representation, graph layouts in information visualization, and vector field topology analysis.

TBA

TBA

TBA

TBA

Joint work with H. Carr and T. Moller

Many data sets are sampled on regular lattices in two, three or more dimensions, and recent work has shown that statistical properties of these data sets must take into account the continuity of the underlying physical phenomena. However, the effects of quantization on the statistics have not yet been accounted for, nor has the impact of the choice of lattice. This paper therefore reconciles the previous papers to the underlying mathematical theory, develops a mathematical model of quantized statistics of continuous functions, and proves convergence of geometric approximations to continuous statistics for regular sampling lattices.

TBA

Rhys will report back on his experiences at EuroVis 2010 in Bordeaux, France.

TBA

This talk will present the ongoing research at the ThinkLab on collaborative workspaces. The first part of the talk will focus on the outcome of the CoSpaces project which investigated the design and implementation of a collaborative platform for the aerospace, automotive and construction sectors. The requirements for the system were gathered by analyzing current collaborative practices and future visions. This project resulted in creating three types of workspaces : co-located, distributed and mobile. The second part of the talk will focus on the collaborative platform that is being developed for supporting various stakeholders in urban regeneration. Building on previous experience with the Black Country Urban Regeneration programme, current work is now focusing on Salford City to provide a comprehensive urban simulation framework.

TBA

Large graphs, both static and dynamic, are present in numerous applications. As data set size increases, drawing the entire graph, that is giving positions to all nodes, becomes computationally expensive and resultant drawings are often difficult to understand. Thus, interactive graph visualization systems may help users understand this data.

This talk presents a variety of techniques for static and dynamic graph visualization. It also presents two user studies which, in part, validate some of the design principals behind these systems. All of the systems presented leverage either the topological structure or attributes associated with the nodes and edges of the graph to create summaries of higher-level graph structure.

Rebecca will share her experience at CHI 2010 in Atlanta, USA.

A nested model for visualization design and validation
by Tamara Munzner
IEEE Transactions on Visualization and Computer, 15(6), pp 921-928, 2009.

Searching information on the internet is popular but tiring. Searching for a previously visited web page is even more painful. At the first stage of my PhD, I have focused on investigating how people search specific information on the internet, why they need to find web pages again and how they do that. Today, some web history tools have been developed to support users in navigating in their own web browsing history. From a literature review of these tools, I have determined 3 main issues when developing a future tool: (1) what data about an individual's web browsing history should be captured, (2) how data should be processed and (3) presented to users. I will present some reviews and bring the first issue for discussion by the group.

Visual representations of heap memory graphs can aid in debugging and help to develop program understanding. However, such structures may contain thousands of objects and, while some drawing conventions exist, have no obvious spatial organisation. For the user, navigating these graphs can be challenging, especially where their own mental map of the data space is incomplete.

To ameliorate this problem we employ a systematic approach grounded in the principles of navigational landmarks. We identify sub-graphs within the heap that correspond to significant design abstractions and use various techniques to highlight and organise these structures. Thus, the user is presented with recognizable features that are linked to other representations of the software. We claim that the enhanced representation can support also existing debugging tools by providing the user with a usable "map" of an otherwise abstract data space.

The results are demonstrated using an instrumented version of the Visualization Tool Kit (VTK), a widely-used architecture for data visualization.

With Tommy Hinks & Debra Laefer (UCD)

One of the core problems in urban planning is the acquisition of geometric models of individual buildings. Feasible on a small scale with surveying techniques or terrestrial-based LIDAR, it becomes increasingly difficult and expensive at large scales. And, while airborne LIDAR is common for terrain surveys, facade details on vertical elements pose particular problems for aerial scans of urban environments. We therefore describe how to plan a flight path to maximize the quality of aerial LIDAR point clouds for urban modelling.

Clustered heatmaps and profile plots are common techniques to visualize collections of gene expression profiles from high-throughput microarray studies. However, these techniques do not scale well for data sets where gene expression levels have been measured under hundreds or thousands of experimental conditions.

In my talk I will identify the problems of visualizing expression profiles with a very large number of conditions and introduce the Space Maps method, which addresses these problems. Space Maps are 2-dimensional, multi-resolution glyphs that represent gene expression profiles. Their arrangement in 2-dimensional space represents relationships between the gene expression profiles.

http://www.gehlenborg.com/research/spacemaps

The first 6 months of my PhD has focused on analysing the problems that occur in collaborating between a Powerwall and other, less able displays. These problems will be analysed as part of a case study that has been developed through collaboration with the centre for integrated Petroleum Engineering and Geoscience (CiPEG). I will present some potential solutions to these problems for discussion by the group.

15:00 FLEGG, Rob, Title: Visualizing traffic delays in real time. Opening with an overview of the project and description of the data feed, the visual encoding techniques used for the display will then be discussed. Firstly showing the encoding preference originally chosen and then describing the results of the visualisation questionnaire and the resulting changes to the encoding. The plans for analytical tools will then be briefly described before opening the floor for questions.

15:20 TERZZA, Michael, Title: Eye Tracking with X-Ray Images.

• Brief outline of what DEXA images are, why they may be helpful, why they're not being utilised to their full potential.
• Description of how & why we are using eye tracking with the radiologists
• Outline of the eye-tracking setup and software developed
• Explanation of possible use of the collected data to develop either:
• Learning software / systems for junior radiographers.
• Software to aid in the viewing of DEXA images whilst radiologists are reporting on them.
• My ideas/thoughts on both of the above
• For Learning:-
• Have a "follow me" approach for radiographers to follow the gaze of previously recorded radiologist's views.
• Show heat-maps of areas covered (based on model answers)
• Show scan-paths of gaze (based on model answers)
• For Viewing:-
• Best opacity of overlayed images
• Possibly variable opacity
• Edge detection of bony image overlayed
• Grid overlayed onto soft-tissue image
• Other:-
• Use Algorithms for clustering fixations
• Categorise scan-path/fixation data
• Can anyone think of anything better?

15:00 GARLICK, Thomas, TBA.

15:20 GUDYNAS, Mantas, Title: Automated Generation and Applications of Metro Style Maps. Metro style maps are widely used in areas where data visualisation is required. Manual drawing of them introduces number of challenges due to number of criteria for positioning nodes. This presentation will be focused on automated generation of metro style maps as well as real world applications of them.

15:40 JONES, Scott, Title: CUDA: The Technology Today. It is a look at peoples usage of cuda programming, and the kind of problems they can be used to solve, It will also include a brief overview of pyCuda, a python wrapper around the C API, with an example of someone thats used it.

3pm CHALLIS, Steven Title: Decoupling Illumination from Isosurface Generation Interactively visualising scalar datasets with global illumination is a difficult task due to the computational requirements of physically-based renderers. The simple solution is to extract and illuminate a set of isosurfaces in order to precompute an illumination grid for use as a 3D texture map. I will present a new method for generating the illumination grid which overcomes the problem of non-uniform sampling associated with the repeated extraction and illumination of isosurfaces.

3.20pm BIGGS, Andrew Title: A visual approach to Enterprise Search. An overview of what Enterprise Search is, motivation for doing it (and doing so better) and some of the problems associated with it. Then a summary of the findings from my user survey and my thoughts on where that could lead a new tool.

3.40pm TSENOV, Krasimir Title: Ray-Tracing using Parallel Haskell I will (try to) talk about what ray-tracing is and show an example of a good ray-tracer, then I will talk about how it can be parallelised and why Haskell is a good language to write parallel programs with.

I'll present examples and a framework, which might help us (not just me!) answer that question. In particular, I'll focus on interaction as an essential part of visual analytics applications.

A presentation on the first 6 months of my PhD project, investigating concepts of search and attempting to represent search problems and algorithms clearly in Haskell. The talk covers Local Search, Haskell, some of the problems that have been encountered, some of the approaches that have been tried and the current ideas that are being investigated.

Augmented Reality (AR) is an emerging technology that allows computer graphics to be superimposed onto real world scenes at interactive frame rates. Many applications are currently in development both for work and pleasure. In this talk the development of the Bangor Augmented Reality Education Tool for Anatomy (BARETA) will be described. BARETA adds computer-generated content onto a video stream of the real-world, registered with anatomically correct anatomy models produced from patient specific data.

The talk will discuss some recent work in a KTP project with Nominet (UK) whihc set out to examine behaviour of users of the WHOIS service, essentially be analysing log files. Some simple visualization techniques yielded valuable insights. We also looked at the concept of typo-squatting domain names and how this can be characterised. Experience of this project has led to some reflection on the ease of constructing visualizations and of recording how visualizations are used. The talk will look at some work in this area too, and parallels with the fields of digital curation and digital forensics.

The talk will conclude with some examples of visualization applied to the Web of Data, and issues that are arising from this.

The overall aim of the talk is to stimulate thoughts and discussion.

Slides from the talk available here

Digital image compositing is the layering together of image components to make a novel final image. It has been used for many years and is widely used for Hollywood and TV visual effects. The core technique is an effective but ad hoc solution. It turns out to be very closely related to projective geometry which, in its practical form, is the basis of perspective and other projections used by computer graphics for even longer. One consequence of this is that the mathematics, the graphics processing pipeline and the associated tools can be substantially the same for both colour and geometry.

The talk will describe the connection and will then go on to show some more sophistated image modifications in which geometry and colour influence each other. It turns out that there are two separate occasions when this work at Bath was influenced by the University of Leeds ...

This is likely to be my final talk before I finish my time at Leeds, and I will be filling you in on the work I have been doing the past year. At my last VVR talk, I was trying to classify both data and visualisation. This was for the purpose of developing an interactive visualisation system that was designed to work with Powerwall displays and take advantage of the real estate available. I will talk about the case study data set that I was given to me by the Institute for Transport Studies, and the types of visualisations they use to analyse their data. I will also be demonstrating the visualisation system I developed, as well as showing some multi-window interaction techniques. I will conclude with the evaluation of the system, which was done by observing six MSc students use the wall for a total of 12 hours (over four days).

This talk will describe work aimed at the combined problem of collection, management, visualisation, search and analysis of large amounts of complex uncertain data. Our aim has been to deliver value to users from large and complex data sets, based on our background in neural network based pattern analysis methods. Our initial focus was in the engineering field undertaken in the DAME project on the analysis of Rolls-Royce engine data. That work developed a distributed search engine (PMC) and a signal data search and analysis system (Signal Data Explorer) which has now been commercialised by Cybula and sold to customers such as Bombardier (for rail) and Scottish and Southern Energy. More recently we have transferred the methods into the scientific medical field within the £4.5M CARMEN project, which has explored the problems of distributing analytical services and data within a distributed computing infrastructure. The project has just launched the CARMEN portal (www.carmen.org.uk) for use by neuroscientists to share and analyse electrophysiology data. The talk will describe the challenges faced by the end users, and how the methods developed under these projects are helping them, and what issues remain to be addressed.

Information spaces such the WWW are the most challenging type of space that many people navigate during everyday life. Unlike the real world, there are no effective maps of information spaces, so people are forced to rely on search engines which are only suited to some types of retrieval task. As a result, knowledge workers spent a substantial amount of time trying unsuccessfully to find information that they know exists. The present paper describes a new method for creating maps of large information spaces. The maps use a city metaphor, comprised of streets (branch nodes) and buildings (leaf nodes), which are generated entirely automatically from webcrawl data. A technical evaluation was carried out using data from 112 universities, which had up to 485,775 pages on their websites. The effect of different layout rules on INSIGHT maps is presented, showing how they are more compact than radial graphs such as the Bubble Tree

Sophisticated methods of visualisation allow a user to explore datasets and gain insight from the information being displayed, but are often difficult for non expert users interact with. One such method is volume rendering which creates composite images using a transfer function which maps different optical properties to a given input. Previous work used the contour tree, which represents topological features of a dataset as branches of a tree, to apply different transfer functions to different aspects of the data. While this allows for individual features to be highlighted and explored, it suffers from a usability standpoint. The user is required to have knowledge of the mechanics of the contour tree, instead of using the data structure purely as an abstraction of the data. We therefore propose to combine the principle of this previous work with the idea of a thumbnail gallery and the hierarchy of topological features to give non expert users a simpler interface.

Hamish returns from VisWeek 20009 in Atlantic City to give us a report on the conference.

Most previous research into the effect of body-based (proprioceptive & vestibular) information on navigation has used basic spatial tasks such as path integration. However, key findings from that work (specifically, the importance of rotational body-based cues) do not seem to apply to "everyday" tasks performed in large-scale spaces. I will describe my recent research that has investigated the effect of rotational vs. translational body-based information when participants learned routes and developed survey knowledge in large-scale virtual worlds. Some of these worlds were shrunk to fit within a 12 x 13 meter tracking laboratory, but in the full-sized versions of the worlds participants walked hundreds of metres on an omni-directional treadmill. The most important factor was proprioceptive cues for translationary movement, which produced a large, significant improvement in the rate at which participants developed both route and survey knowledge.

- Welcome new members and introduce ourselves;

- Discuss role and operation of the VVR group meetings

- Discuss strategic issues