A number of students have asked us questions like “What’s an abstract?” and “What kind of work gets in?”. So we asked former attendees if they’d be willing to share their abstracts.This page has a selection from across contests (first year, second year, finalist and MSc) and from different Lovelace years (2012 was in Bath, 2013 in Nottingham, 2014 in Reading and 2015 in Edinburgh).
All of these abstracts were finalists – that is, all of the abstracts featured on this page were successful submissions, and the students in question got to come to the event with their poster and we re-funded all their travel.
First year poster abstracts
“Democracy 2.0”: The potential effects of technology on political participation and democracy
Poppie Simmonds (University of Birmingham) 2014
Recently, social media has been an important tool for countries who are establishing democracy, by enabling communication between groups to create change. As witnessed in the events of the “Arab Spring”, technology as a force for revolution in democracy is powerful and contagious. The ability to access information freely, organise meetings between groups, and document events from the ground to gain support were some of the key factors in the successes of the political movements in the Middle East and other countries. In many other places around the world however, democracy is already a well established tradition, with systems being engrained in history. Whilst countries such as the UK are at the stage where adults have the legal right to vote, this is often taken for granted and political apathy is rife, so technology could shape the way we engage with democracy and create a culture of political awareness in established democracies, especially with the traditionally disengaged younger generation. The creation of e-democracy through introduction of systems like e-voting could bring a more secure and convenient way to take part in elections to the masses. Using e-democracy to improve age old democratic systems may be the way to inspire more people to care about politics and take an active role, which is extremely important. It has been seen that social media and other technologies have already had a huge impact in very recent history, but the potential for future advances to change democracy for the better is exciting.
Second year poster abstracts
Catherine De Roure and Sophie Drake (University of Bath) 2012
Music is a concept that follows many key structures and rules that you would commonly come across when dealing with mathematical theories. It’s been known that there is a strong correlation between the subjects of maths and music. Because of its foundations and it’s familiarity it’s become a key interest in research across the board, and when mixed with ambitions of computer science, it has provided us with a lot of interesting potential, to an extent that may have not even be realised yet.
We’re going to discuss the mathematical principles that make up the fundamentals of music and sound, and thus go on to cover the current implementations of the research going on now. We will also be touching on the directions in which this research is heading, and the ideas that have come from this.
It’s not all “Boring” “Maths” and “Typing”
Charlotte Godley (University of Hull) 2013
The three words young people, particularly women, use to describe Computing within school tend to be “boring” “maths” and “typing”. Whilst ICT is in the process of being improved, this suggests that ICT and CS are being undersold within high schools and it needs to be shown the multitude of areas that the subject involves, proving computer science is so much more than these three words. The poster covers areas such as physical computing and sensing with the raspberry pi, arguably the most accessible and well-advertised community in which hobbyists, students and experts can directly program changes to hardware and take in input from their environment without requiring a laboratory or extortionate funding, the accessibility and ease of publishing games with XNA which has recently been scrapped by Microsoft but is still doable via means of the MonoGame project, and various uses of computer science in medicine and science. Whatever a person’s interest, there will be an area of Computer Science which will take a student’s interest, and this poster seeks to portray that Computer Science is anything but boring and should not be portrayed in the light of typing in a word processor.
Augmented Reality – what future can it have on campus?
Roseanna McMahon (University of Bath) 2014
Over the past few years, a variety of companies have branched out into introducing a selection of augmented reality (AR) ideas into their applications and products. In this presentation I aim to consider these and discover if there is a place for augmented reality ideas on campus, whether it be as part of our university education or as part of the wider university community. AR is a real-time copy of reality that is placed on top of reality through some user device, for example a smart phone. Using the device, the user can experience live additions to actuality and manipulate and interact with the digital world in front of them on their screen. Common uses recently have been navigation, tourism and entertainment but the first talk of AR can actually be traced back 100 years to a novel by Lyman Frank Baum. This famous author of the acclaimed “The Wonderful Wizard of Oz” mentioned a glasses-type-device called a “Character Marker” in his book “The Master Key”. It allows the wearer to see a character description on other humans’ foreheads on meeting them, though he included a comment on how this was likely an idea centuries ahead of its time. Nowadays we are seeing commercial uses of AR, first on the YELP smart-phone app, as a restaurant-finder, and now as a recently announced app available for the new Google Glass. The rest of this poster should discuss how these ideas can be adapted to use on campus and how they can benefit and support our university experience.
Could a robot make this poster?
Milka Horozova (Queen Mary University of London) 2015
Advances in the field of Artificial Intelligence (AI) have found practical applications in numerous areas of human life, from self-driving cars and search engines, to face recognition and programs that learn by themselves when only given a goal. Robots that paint, compose music, write poems, and have their output indistinguishable from human art, challenge our definition of creativity. These and other examples of successfully implemented AI techniques prove that machines can mimic and even outperform people in domains long considered humans- only territory. This poster presents the ways humans and machines learn, and elaborates on the differences and similarities between natural and artificial intelligence. It examines some of the major machine learning paradigms, and extends on the way artificial neural networks algorithms are inspired by the human brain. Finally, the poster explores the future possibilities and potential benefits of AI, and examines the extent to which humans and robots could compete in the economy of the future.
Third year poster abstracts
Data Mining of Horse Racing Results
Maithreyi Venkatesh (University of Edinburgh) 2014
Horse racing is a multi-billion pound industry in Europe and is one that is primarily funded by betting. This has led to the research and development of wagering systems which employ a number of different techniques to predict the outcome of a horse race. In some cases, they use an optimal wagering strategy according to those results. These wagering systems are made up of two distinct algorithms; the first is an algorithm which predicts the ‘most likely’ winner of a race and the second is a betting algorithm. The output of the first algorithm, which models the horse race process, becomes an input to the betting algorithm which uses these probabilities to advise a bettor on how much they should bet on a particular horse.m w Regression analysis is one of the models used by these wagering systems to predict the place of a horse in a race. This is the process of predicting a numeric output, or dependant variable, using a set of corresponding input variables, or independent variables. In data mining, regression analysis provides techniques for analysing how value of the dependent variable changes based on the changes made to the set of independent variables. Regression techniques work naturally with numeric attribute hence it is suited to be used in the prediction of the performance of a horse as it provides flexibility with regards to what it can be used to predict. This project aims to predict the speed of a horse using Regression analysis. The dataset being considered will only include the races which took place in the UK and Ireland.
Sexualisation, Objectification & an Invisible Audience: Female Portrayal in the Video Game Community
Michelle Brown (De Montfort University) 2015
Female representation within video games is a controversial issue. Female characters, such as Lara Croft, are often sexualised. Attitudes, demeanours, attributes and clothing all contribute to the overall representation of women in video games, which inspired a
study on how gamers would like women to be portrayed. More than 150 responses (52% female; 47% male) showed that around 92% of people agreed that women are sexualised within video games. When asked how they would like to see women represented, the majority used words such as independent, outgoing, strong-willed, intelligent, brave, heroic and normal, which, in contrast to how women are portrayed in video games currently, shows that the video game industry has a significant misconception about what might “appeal to the primary users of the product” (Scharrer, 2004). When shown three images of the same female character in three different items of clothing (revealing, covered, and in between), 84% of respondents preferred the more covered version; only 16% preferred the image with the least amount of clothing, with most reasons being that they wanted women in “sexy” clothing. Numerous respondents also mentioned that they felt female gamers aren’t the target audience for games, which is a serious issue considering that women actually make up around 45% of the video game community (ESA, 2013). It’s clear to see that the video game industry needs to listen to their complete audience as the gender gap is rapidly closing, and improve the representation of women in their games. ESA. (2013). ‘Essential Facts About The Computer And Video Game Industry’, paper presented at Electronic Entertainment Expo (E3), Los Angeles, 11-13 June 2013. SCHARRER, E. (2004). ‘Virtual Violence: Gender and Aggression in Video Games Advertisements’. Mass Communication and Society, 7(4), p. 393-412.
Saliency Driven Monocular Path-finding for Future Autonomous Robots
Bruna Pearson (Durham University) 2015
For a robot to autonomously navigate in unknown environments, it requires a map indicating traversable scene regions for robot motion. The key challenge when building a real-time traversability map is to automatically differentiate between obstacles and path boundaries within the scene. Additionally, the efficiency and accuracy of the traversability map depend on the ability of the algorithm to adapt to variable environmental conditions and illumination without increasing either the complexity or the computation that degrades real-time performance. This research compares the accuracy of the traversability map when using a visual saliency technique either as an input channel or a substitute for the current edge detection filter within prior work in the field. It also compares the enhancement of the image segmentation process when using different colourspace. The experiments were realised on a Pioneer 3-AT robot platform, which executes commands using a subsumption architecture. Each layer in the subsumption architecture operates individually towards a goal, while interacting with other layers. Through these interactions, high-level layers can instruct lower-level layers to perform motion. In this setup, motion is controlled by inputs originating from the traversability map, which provides a high level set of coordinates, while built-in sensors provide data for the obstacle detection and avoidance sub-level of motion control. Our results demonstrate the accuracy for pathway detection by comparing the traversability map output with a ground truth and by evaluating the navigability approach used for monocular guidance in mobile robots.
MSc poster abstracts
Trust Metrics for Online Social Media
Kaushalya Kularatnam (University of Oxford) 2014
The proliferation of Internet usage has resulted in huge amounts of information being made available online. A big contributor to this are social media networks such as Facebook, Twitter and blogs which act as public boards of discussion. However, whilst the scale and variety of data has considerable value for commercial and social purposes, the quality and validity is often questionable. Opinions are frequently posted to social media sites and are often a mixture of fact, speculation or rumour whereas user-driven sites such as Wikipedia are often questioned for their trustworthiness. Information trust and quality metrics can be used to address the issues of validity and quality through the use of automated assessments. These metrics would be used to identify different aspects of an informational source and look for independent verification of the quality of the data. The metrics would need to be tested against real-world examples and be context-dependent; the nuances of language can change the meaning of a piece of information drastically through the use of slightly different words. The aim of this work is to provide a reliable and repeatable method of verifying information on the internet independently and create a more trustworthy online experience.
Seeing Through Walls: Handling Large Datasets
Maitreyee Wairagkar (Univesrity of Reading) 2014
Terahertz (THz) radiation consists of electromagnetic waves with frequencies between 100 GHz and 10 THz. THz spectroscopy deploys this radiation range to scan various materials for medical imaging and other applications. THz radiation penetrates up to 1 cm of optically opaque materials including wall plaster and is a non-contact, non-destructive technique used for analysing cultural heritage, specifically imaging mural paintings that have been plastered over. Seeing Through Walls project uses THz imaging to scan the paintings on the old walls of churches and caves hidden behind plaster, paint or other mineral deposits and recreate them digitally. Terahertz images comprise huge amounts of data. Each pixel is scanned by separate time series signal making data manipulation slow. This project developed an efficient database system to store and manage the 3D datasets and integrate them with customised image visualisation software developed in JAVA capable of reconstructing the images in real time. The project has significantly increased the speed and efficiency of reconstruction of THz images of historic wall paintings by creating a database system in MySQL to organize and systematically store the data. The developed visualisation software allowed retrieval of data from data-handling system and recreation of images in real time by creating blocks for manipulation in frequency and temporal domain. The user friendly GUI of the software provided functionality to store raw data automatically in the database and view multiple reconstructed image blocks at on the same time on single screen to create an entire scanned picture.
Automating Repetitive Tasks
Maithreyi Venkatesh (University College London) 2015
Users spend a great deal of time manually performing repetitive tasks that they should automate. They choose to manually type out commands employing more and more keystrokes rather than writing out a program once and saving keystrokes. One reasons for
this is a lack of knowledge of how to automate tasks. Another is simply that users underestimate the number of time they will repeat a task while simultaneously overestimating the cost of automation. Our aim is to save users keystrokes which we employ as the measure of effort put in by the user to carry out a task.We present a method to detect the repetitive tasks and automatically automate them on behalf of the user. This saves the user the effort of having to learn how to automate tasks and the effort of having to do so thus saving them keystrokes. This ultimately increases the payoff to the user. We apply this method to a command line tool that automatically detects and generates aliases and mini-scripts based on the user’s bash history file. To do this, we use language models and manual analysis of the bash history files to identify and create templates for various tasks performed by users. Following this we use an ML/NLP algorithm that learns from the bash history to detect and generate aliases and mini-scripts.