Research Activities

Feb 2012

CISA Seminar Talk: Efficient Management of Compute and Memory Intensive Processes on Large-scale Videos using Multi-core Environment.
Collaboration with NCHC: Installing Open Grid Scheduler on multi-core VMs.
Collaboration with CWI: Integrating workflow and user interface components.

Jan 2012 Paper submission and reviews for KES-AMSTA 2012

Paper submitted: Goal, Video Description and Capability Ontologies for Fish4Knowledge Domain.
Reviewed three papers.

Dec 2011 F4K's First annual review meeting, Catania, Italy

Presented project ontologies and integration with UN FAO's Network of Fisheries Ontologies.
Liaised with NCHC, Taiwan on setting up the VM servers, clusters and supercomputer for F4K's use.

September 2011 ISPA Conference, Dubrovnik, Croatia

Presented talk on paper: A Flexible System for Automated Composition of Intelligent Video Analysis.

September 2011 Work visit Catania, Italy

Collaborated with image processing experts on modularising IP detection and tracking tasks.

August 2011 F4K Project meeting, Edinburgh

Discussed ontology work in progress and initial thoughts on parallelisation of IP tasks on HPC machines.

June 2011 KES-AMSTA Conference, Manchester

Presented paper at Intelligent Workflow, Cloud Computing and Systems special session.

May 2011

DIR-group talk.

April 2011

F4K 2nd project meeting, Taiwan.

April 2011

3DPAS Workshop.

March 2011

CISA Seminar talk.

November 2010

F4K 1st project meeting, Edinburgh.

August 2010

Passed Viva!

July 2010

Thesis submitted!

January 2009 - present

Workflow development, collaboration with C.Spampinato on implementation of agreed independent executables for video classification, fish detection and counting task.

September - December 2008

Visiting University of Catania, Sicily.

August 2008

Coding and preparation for Italy.

July 2008

1) Paper-writing: IGI chapter, workshop paper.
2) Back to coding.

June 2008

1) OpenKnowledge work.
2) Paper-writing: IGI, EKAW workshop.

May 22nd Phone Meeting Edinburgh-Catania (Italy)

Also Present: Concetto, Jessica, Bob.
Stepped through classification tasks performed by Concetto. Discussed the possibility of generating scripts for my use at the workflow end of things.

Apr 15th - present

Coding...

Apr 14th IGI chapter deadline

Mar 20th Teaching ends

Jan - Feb 08

TODO:
1) IGI chapter preparation.
2) OpenCV compilation.
3) Euro-Par paper review.

December 10th - Jan 2nd Break

Submitted review for CCGrid 08.
Book chapter proposal accepted.

November 30th

Prepared proposal for IGI book chapter, work with Areti and Jessica.

Oct 31st - PhD Proposal talk

Feedback from Panel:
1) Ontologies should be built based on agreed principles.
2) Model the tasks as activities and a set of constraints, if this is compatible to planning.
3) Perhaps plan refinement from a detailed plan.
4) Find best modules for a combination of constraints based on requirements.

Oct 29th - Meeting with Jessica and Bob

Went through slides for PhD talk on Wednesday for feedback.

Oct 23rd - CISA Seminar Planning Talk by Hector Geffner

Oct 12th Meeting with Jessica

Discussed ideas for data model and linking planner with ontologies using a data language.

Oct 5th - Meeting with Jessica and Bob

Discussed more on implementation details. Should concentrate on plans for specific scenarios, ie how to tie the ontologies, planner, libraries together. Attempted to call appropriate IP scripts within WF engine.

Sept 20th - 27th ICAPS 07, Providence, RI

Sept 14th - Discussion with Concetto

Final discussion with Concetto on VP tasks, files and future collaboration before he leaves.

Sept 7th - Meeting with Jessica

Discussed ideas for planner - needs to be as generic as possible, process models for tasks that are hard-coded for specific tasks are most likely to be in workflow. But if they are breakdown of methods they could be encoded as HTN tasks.

Sept 4th - Task Breakdown

Derived the composition of tasks, need to improve on recursion and conditionals.

August 25th - 29th CAIP Conference, Vienna

August 8th - 17th Process models

Worked on deriving the process models for the partial plans.

August 7th - meeting with Concetto

Went through all the steps taken in the video classification task, fish detection and tracking conducted by Concetto (IP expert).

August 2nd - meeting with Jessica, Bob and Concetto

1) Feedback on PhD hypothesis from Bob.
2) Discussion on future work with Concetto.

July 27th - meeting with Jessica

Discussed the Planning mechanism with the ontologies, process and case libraries.

July 18th - meeting with Jessica and Concetto

Further refinement of Prolog file.
TODO - Planner.

July 12th - 15th break

July 11th - Prolog file output

Discussed clip and frame properties for Prolog file.
Identified different types of fish.
Added colour information such as HSV and also presence of noise.
Discussed criteria for interesting frames.

July 2nd - 6th working with Concetto

First stage video classification tasks done succesfully.
Fish detection in progress and near completion.
Fish tracking and counting also in progress, to be improved.
Fish classification as next step.

June 21st - meeting with Concetto, Jessica and Bob

Presentations and discussion on related VP tasks.

June 14th - meeting with Jessica and Bob

Discussed presentation for Concetto (summer visiting researcher) next week.

June 6th - next meetings

Thur June 14th 2-4pm with Jessica and Bob.
Thur June 21st 2-4pm with Jessica, Bob and Concetto.

June 4th - meeting with Jessica

1) Continue with Planning, representation language for problem domain.
2) High level goals to be captured from image processing script(s).
3) Prepare presentation for June 21st to give background and problems we are trying to solve.

May 18th - meeting with Jessica

1) Advice on Planning approach.
2) Find heuristics on IP steps taken.
3) TODO: Report on work done with vision libraries.

May 10th - 31st Planning

Worked on Planning basics, Planning in Grid workflows, etc.

May 9th - meeting with Jessica

1) Submitted proposal draft, to be proof-read.
2) Discussed prospects of collaboration with IP experts.
3) TODO: work on Planning for Workflow.

April 25th CISA seminar - Simon Buckingham Shum (KMI)

The way research discourse is done is more digitally mediated. Presented some tools that provide visualisation and the capabilities that capture the way this could be done.

April 23rd - 27th IP library

Installed and worked with Pandore library for simple detection.

April 16th - Meeting with Jessica

Brushed up more specific ideas for PhD proposal.

Mar 19th - Apr 4th CAIP paper

Collaboration with A.Renouf (Laboratoire GREYC, Equipe Image) on ontology-based IP solutions for video analysis.

Mar 15th - meeting with Jessica and James

Discussed progress with hypothesis generation, methodology, evaluation.

Mar 12th - meeting with Bob and Jessica

Formulate hypothesis in workflow problem and verify later if ecogrid data could be an ideal domain for it.

Mar 7th - Meeting with Jessica and James

Feb 19th - IVR Matlab demo

Step by step solution for first practical.

Feb 16th - meeting with Jessica

More discussion on WF QoS. Focus on Image Processing tasks for now.

Feb 12th - OpenCV

Installed and ran simple program using OpenCV 1.0 on Linux.

Feb 11th - more on Video analysis and WF QoS

Looked up Matsuyama (1989) and Cardoso (2004).

Feb 8th - meeting with Jessica

Discussed QoS for Workflow in Grid, differences between workflow process selection performance and workflow grid node performance were made clear.

Feb 5th - meeting with Bob and Jessica

With the problem of tracking down Jim Gray using vast number of satellite images, some interesting discussion was made. And stressfully the direction and focus of my work must be determined.

Jan 31st - Reading on Video Understanding Systems

Some related work, in particular literature review for knowledge-assisted video analysis effort.

Jan 29th - meeting with Bob, Jessica and James

Got some feedback on how to go about populating the ontologies. Some vision assistance is required if a system with high performance is required.
TODO:
1) determine the input and output parameters of capabilities and that they do not cause conflicts.
2) work on available tools/techniques for further feedback next week.
3) suitable video format.

Jan 23rd - e-Science Talk Jessie Kennedy

"Exploiting Diverse Sources of Scientific Data". This mostly deals with the fact that data will continue to get overwhelmingly large and how present efforts (metadata, ontologies, LSIDs, WF tools, Grid/Web technologies) must evolve to deal with it. Focusing on the first three, concepts such as meta-utopia, reusing and convering to Life Science Identifiers (LSIDs) are further discussed using examples from biological data. The use of tools and technologies will go a long way to help speed things up.

Jan 18th - Meeting

TODO:
1) Email Bob Fisher to meet to discuss ontologies and walkthrough.
2) Work on prototype - manual walkthrough.

Jan 16th - Meeting

IET Software accepted. To send final version.

Dec 18th - Dec 21st WI/IAT/ICDM 2006, Hong Kong

December 18th - International Workshop on Service Composition talk.
December 20th - Web Intelligence Conference talk.

Dec 12th - Jan 13th Break

Dec 8th - Meeting & Submission

1) PhD meeting, to manually run system as laid out by walkthrough.
2) Submitted ontoMapping, to sign copyright still.

Dec 5th - ontoMapping Draft

Managed to write paper up, for further comments.

Dec 1st - Walkthrough II

Worked a bit more on ontologies and the walkthrough.

Nov 28th - Meeting

Discussed major revisions for ontoMapping. Deadline Dec 9th.

Nov 20th - Meeting

TODO
1) Do a manual step-by-step implementation of walkthrough for simple VP tasks.
2) Major revision for ontoMapping paper.

Nov 13th - Meeting

Further refined and separated ontologies.

Nov 6-11 Ontologies and VP

Obtained description on existing IP ontology and 2 relevant papers.
Building separate ontologies and working on VP basics.

Nov 3rd - meeting

More discussion on VP ontology and system's workings.

Oct 27th - meeting with Jessica and James

Discussed VP tools ontology.

Oct 25th

1) Done with KMM tools.
2) Moving on to VP ontology, identified a few existing ones.
3) Identified the following libraries:
i) OpenCV
ii) LTI (C++)
iii) HTK (HMM classifier)
iv) Lib SVM
v) ITK (medical imaging)
vi) Khoros
vii) STL (C++)
viii) Pandore
ix) Vigra
x) Limp
xi) JAI

Oct 18-20 Break (Firbush)

Oct 16th

1) Done with draft of KMM tools.
2) Done with distribution of IRR reading list.

Oct 11th KMM survey

Finally got started on a much delayed and owed work. Hope to finish it by end of the week.
Other progress: attending IVR course, reading Fundamentals of Digital Image Processing.
TODO: VP ontology + process models and the libraries.

Oct 3rd SSP Talk

Talk in CISA. Slides [PDF]

Sept 18-29th WI/IAT Workshop SERCOMP06 paper

Revise workshop paper to be more technical, i.e. more Grid services-oriented while conference paper is left to be more general and workflow-oriented.

Sept 12-13th IDIAP 15th Anniversary Workshop

Gave talk and gained insight into some vision aspects, especially with regards the specialised techniques as opposed to the generic functionalities that we seek. Image enhancement, segmentation and classification should be looked into more detail which involves more low-level processing.

Sept 4th - TODO

1) Prepare IDIAP slides.
2) KMM survey.
3) WI short paper acceptance.

August 21st-Sept 3rd - Break

August 16th - meeting with James

1) IP tools ontology building - 2 papers given.
2) Conduct small exercise using matlab IP toolbox on segmented video clip.
3) Prepare talk for IDIAP workshop next month.

August 15th

Submitted IAT'06 review.
Submitted onto mapping paper to IEE Proc. Software.

August 11th

To work on next step - build ontology of video processing tools.
Identified two more relevant papers.

August 6th

Submitted to Service Composition Workshop.
Withdrew paper from Semantic E-Science Workshop.

August 4th

1) Pretty much done with onto mapping paper, just need to be proof-read.
2) Planning on submitting to WI workshop and if so, withdrawing from Semantic E-Science workshop.

July 28th - meeting with Jessica

1) Discussed onto mapping paper.
2) Talk for IDIAP workshop acceptance.

July 19th - meeting with Jessica

1) Discussed improvement on ontology mapping paper.
2) Next step for phd work.

July 17th - submission

Submitted to 2 more workshops, e-Science and Semantic e-Science.

July 11th-12th - submission

Submitted to IDIAP workshop and WI conference.
Also need to work on Ontology Mapping paper for ITSSA.

July 5th - paper writing

Working on improving the draft from last week.
Deadlines July 11th, 12th, 17th.

June 27th - draft submitted

Sent first draft of paper to be proof-read by Jessica and James.
Things yet to be done:
1) Finish up KMM tools survey.
2) Submit claims.

June 20th - meeting with Jessica and James

Discussed our work in more detail and what I will be expected to do.
Work on the paper and get feedback before submitting.

June 16th - papers and deadlines

Found some interesting papers on video annotations and workflow systems for the grid.
Also some deadlines for workshops and conferences.

June 12th - meeting with James

1) Look up papers on video annotation.
2) Look up Matlab image processing toolbox.

June 7th - 9th Informatics Jamboree

Prepared a short talk. [Slide]

June 2nd - Things to do

1) Image/Video processing - basics, steps and outcome.
2) Workflow mechanism - read up papers on existing WFs on Grid to refresh.
3) Theoretical Evaluation on existing WF systems; what are the requirements, what they have done, what they haven't done, their advantages and disadvantages, compare with user requirements and propose a solution!

June 1st - Misc

1) Submitted JSME review.
2) Research on KMM Tools.
3) Looked up more resources for work.

May 30th Transkills Course - Paper Production (Alan Bundy)

Course on publishing workshop, conference, journal papers and PhD theses.

May 29th Workflow tools

Concentrating on existing Workflow Composition tools; Triana, Taverna, Kepler, Pegasus/Condor, Codaw to see how they can be applied to the problem domain.

May 20-25th Break

May 17th - May 20th SETN'06

Attended 4th Hellenic Conference on AI, Crete, Greece.

May 12th Meeting with James Malone

Some insight on how to approach the video/images for the task at hand (either generic or specific) was gained. A generic solution would be preferable but it is also more difficult to achieve.

May 5th - Overview

Outlined an initial overview of framework on paper. Much more work and thought should be put in, without a doubt.
Reading video mining to get some basics on video/image processing.

May 4th - Meeting

Overall architecture of framework to be developed and an outline of work to be done.
KMM tool proposal near completion, final phase of editing.

April 27th - Meeting

Acquired book on video mining.
KMM tool proposal to be completed, draft discussed and modified as needed.

April 24th - KMM tool survey

In addition to the list below:
4) TA work on surveying KMM tools, deadline July 10th.

April 21st - Meeting

1) Prepare slides for setn06 and ccgrid06.
2) Propose KMM tool for next year for a deadline in May.
3) The Phd proposal.

Apr 6th and 7th - How to be an Effective Researcher

Course aimed at PhD students that included process and task management, collaboration, student-supervisor relationship, learning styles and recreational activities.

Mar 25th - Apr 4th Break

Mar 24th - PhD proposal

An attempt to put down on paper a draft of the proposal. To be much improved upon return from break.

Mar 23rd Planning and CBR

Presently looking at CODAW, a tool which incorporates CBR for WF Management. At this point, it is quite feasible to say that my work would encompass Planning and CBR, with the possibility of including ML technique(s).

Mar 21st SSP Talk - Extending LCC

First attendance to this talk, though I could not see a direct relevance to my work, it was useful to know some areas of research that the department is undertaking.

Mar 13th Tools, Planning, CBR, ML

Having done some thorough searches on tools and work on the techniques, there seems to be room for some significant contribution. Some on-going (administrative) preparation for the conference in May.

Feb 24th More Discussion

Identified some papers with work that are interesting and relevant to the problem domain. Some of the tools mentioned could be useful. I should also look into writing up a Principal e-Learning Fund proposal which is due in early May.

Feb 23rd SMIV2

Identified some interesting talks that could be of relevance to my work. There seems to be the notion of Databases (again) when it comes to large scale data and this may need to be taken into consideration at some point.

PhD Thesis Workshop

Though a little premature to be going to such a course, it was an eye-opener to get started on the work, seriously.

Feb 14th Discussion

The more specific areas that need to be looked into are the combination of Image Processing and Workflow (and the Grid). References to the talks presented at the Second Workshop on Scientific Data Mining, Intergration and Visualisation (SMIV2) could also be helpful.

Feb 7 Formulate topic

It's about time to move on to focus on the PhD. I should probe into the problems that could be faced in image processing (seen as a workflow in our context), how and what attributes should be taken into consideration and investigate several different techniques that could be applied to the problem domain. Is there one single technique that would solve it all, or would a hybrid approach be more practical?

Feb 2 CCGrid accepted

Another good news, the work that is a slight extension to the MSc work submitted to the conference above has been accepted.

23 Jan SETN06 accepted

Good news from the Hellenic conference, paper was accepted. Work required to shrink it to 4 pages. A trip to Greece is in order.

Dec 14 - Jan 14 Break

Nov 30 Another paper submission CCGrid

Following the paper which was submitted to SETN06, the conceptual mapping framework was extended to include ontologies and mapping principles from Compendium-to-IX.

Nov 22 Paper submission for SETN06

This paper which summarises my MSc work was submitted to the 4th Hellenic Conference on AI due in May 2006.

Nov 15 Data model Ontologies

The ontologies for the data model and process model of FBPML and OWL-S are to be represented as general and as standard as possible. The IEEE SUMO is a suggested reference. Clarifications as regards instances of classes and relationships from the meeting with Jessica (14/11) were helpful in the development of the ontologies.

Nov 8 Conceptual Mapping and principles

Based on the references to the Compendium-to-IX conceptual mapping and mapping principles, I have opted to take the following approach:
1) One-layered parent-tracking translation
This resembles the way an ontology is represented, i.e. the links are made explicit between a child and its immediate parent, which will allow relations between non-parent-child nodes to be derived from it.

2) The use of two types of links (part-of and is-related to)
In an ontology representation, we will have subclassOf relation, but also make explicit other relations, such as (FBPML Data Model):
class_property(A, P1, P2).
Can be represented in the ontology as:
Class A 'hasClassProperty' P1
Class A 'hasPropertyValue' P2

There are other properties between classes and instances, thus we will have more specific cases of the 'is-related-to' link.

Nov 1-2 Workshop on Spatiotemporal Databases for Geosciences, Biomedical sciences and Physical sciences. (E-Science Institute)

Observing that many large databases do exist in today's scientific applications, and they will continue to grow larger, the debates include how to manage spatial and temporal attributes of these data. The principal question of whether to database or not to database was also brought up. However, as a minority AI community we could not stress the importance of research within knowledge representation and semantic-enhanced methods to overcome some of these pitfalls.