11th PhD Conference – CompConf 2014


By Joe Chrol-Cannon

Last Monday saw our Department gather together again for the 11th Annual Computing Department PhD Conference, or ‘CompConf 2014’. This was a time when first and second year research students could share their developing work in a stress free and friendly environment and a time for those of us who have been around longer to get updated with all the new things that are going on in the Department.

The PhD conference was opened by Prof. Steve Legg, who spoke on the increasing importance of application driven research and how at IBM, research on intelligent systems was being widely applied to solving customer problems and replacing the paradigm of ‘programming the system’ with one more akin to ‘working with it’. Morning speaker Prof. Andrew Randewich first wowed us with the Physics of nuclear explosions and lasers and then explained the computational challenges that large scale physical simulations pose. The solution lies in bigger parallel computers and better computer scientists to re-formulate the simulation code to better take advantage of them. Afternoon speaker Prof. Peter Fleming emphasised the problem formulation aspects of optimisation design. He turns multiple conflicting design criteria and strict constraints into advantages that help guide the process of optimisation that goes beyond simplistic applications of established optimisation algorithms. Robust solutions are given precedence over fitter solutions in the Engineering world.

Prize Winners

Best presentation: Yin Hu impressed with her biological application of machine learning techniques to the tracking of cell growth patterns. Persisting cells are picked out as suspect Mycobacterium tuberculosis. Time-lapse imagery of in vitro cell divisions provided a visual comfort to soften the confusing ‘Kalman filters’ and necessary biological jargon being sent our way.

Best paper: Ran Cheng proposed a significant new algorithm to estimate the structure of the Pareto front and the distribution of the Pareto set in multi-objective optimisation methods. This provides more potential for a designer to select appropriate solutions from the range that the computer produces.

Best research potential: Shirin Enshaeifar inspired excitement by applying her extended signal processing method for accurately predicting stock prices and decoding human brain waves! The potential here is clearly two-fold: make billions in financial speculation or become a mad scientist with mind reading powers… god help us if she achieves both.

Best poster: Anna Vartapetiance caught us off-guard during lunch with her unsettlingly accurate predictions detecting deception in written text. Don’t lie to Anna.

Special Thanks

A well done and a warm thanks to the wonderful Department administrators Sarah and Denise and also to Prof. Jin and Veronica for organising the conference and all the logistics that go with it. Many thanks are also due to Tameera, Ran, Wissam, Shahrazad and Aamo for their kind support. We would like to express our gratitude to the sponsors of this event, namely AI Corporation, Red River Software, IBM, and AWE.

CompConf is an important opportunity when we can strengthen the connections between different research groups and come together as a Department. The support that our sponsors provide for this event is truly appreciated. The website of the conference can be found at http://phdconf.cs.surrey.ac.uk.