London Hopper Colloquium 2024 - Competition for Women Researchers in Computer Science 

Professor Yvonne Rogers FRS

Professor of Interaction Design, director of UCLIC, Deputy Head of UCL Computer Science

Yvonne Rogers is a Professor of Interaction Design, the director of UCLIC and a deputy head of the Computer Science department at UCL. Her research interests are in the areas of ubiquitous computing, interaction design and human-computer interaction.

A central theme of her work is how to design interactive technologies that can enhance life by augmenting and extending everyday, learning and work activities. This involves informing, building and evaluating novel user experiences through designing, implementing and deploying a diversity of technologies.

A current focus of her research is on what human-centred AI means in practice. She is the CTO of LetThink.com, a start-up that is developing a suite of software tools to enable people to think brilliantly.

She is also one of the authors of the definitive textbook on Interaction Design and HCI (now in its 6th edition), that has sold over 300,000 copies worldwide and has been translated into many languages.

In 2024 she was elected as an international member of the National Academy of Engineering. In 2022, she was awarded the ACM SIGCHI Lifetime Research Award; "presented to individuals for outstanding contributions to the study of human-computer interaction".

In the same year, she was awarded the Royal Society Robin Milner Medal for Computer Science and was elected as a Fellow of the Royal Society as "one of the leaders who created the field of Ubiquitous Computing".

She collaborates a lot with industrial partners and was awarded a Microsoft Research Outstanding Collaborator Award. She also received a MRC Suffrage and Science Award (2020) for being one of the leading women in 'mathematics & computing.'

She is also a Fellow of the ACM; a Fellow of the British Computer Society and a Fellow of the ACM's CHI Academy: "an honorary group of individuals who have made substantial contributions to the field of human-computer interaction."

Yvonne joined UCL in September 2011. From 2006-2011, she was professor of HCI in the Computing Department at the Open University, where she set up the Pervasive Interaction Lab.

From 2003-2006, she was a professor in Informatics at Indiana University. Prior to this, she spent 11 years at the former School of Cognitive and Computing Sciences at Sussex University. She has spent sabbaticals at Stanford, Apple, Queensland University, University of Cape Town, University of Melbourne, QUT, UC San Diego and UC Santa Cruz.

Abstract:

Until now, the design and development of computing systems has been largely research and innovation-led. But changes are afoot. There is growing pressure for researchers to become more challenge-led in what they do.

By this is meant carrying out research that addresses societal needs, such as climate change, through developing more sustainable and ethically accountable computing. 

In turn, it means being more mindful of its aspirations and more responsible for the methods used. However, embracing such a paradigm shift is likely to be far more difficult and demanding than ever before - especially against the rise in popularity of Generative AI.

In my talk, I will discuss the new dilemmas and challenges facing those who wish to embark on this new way of framing and doing research.

Ana Garcia-Sanchez

Director Webex Media Engineering UK at Cisco

Ana leads the engineering group responsible for the development, delivery and operations of the Media Services for Webex Meetings. She has been working within consumer electronics industry for more than 20 years: Digital TV (Philips, Sony, Sky) and more recently Video Conferencing (Webex at Cisco). Ana has a master’s degree in Computer Science from the Universidad Politecnica de Valencia in Spain. She has a strong passion for STEM and diversity in the workplace, especially in engineering.

Hristina Palikareva

Software Engineering Technical Lead at Cisco

Hristina Palikareva has been at Cisco for nearly seven years, working on video conferencing solutions both on the cloud (Webex) and on premise (Cisco Meeting Server). Hristina is a holder of a patent on "Integration of Video in Presentation Content within an Online Meeting". Hristina obtained an M.Sc. in Computer Science from Sofia University "St. Kliment Ohridski" in Bulgaria and a Ph.D. in Computer Science from University of Oxford in the UK, where she developed techniques and tools for automated verification of concurrent systems. She also worked for a few years as a Postdoctoral Researcher at Imperial College London in the UK, where she investigated and implemented techniques for automated testing using symbolic execution. Hristina's talk will cover techniques for optimising call placement across highly distributed data centres.

Manon Flageat

PhD student and Teaching Scholar in the Adaptive and Intelligent Robotics Lab in the Department of Computing at Imperial College London

Abstract: Strength in Diversity for Robots

Robots and other decision-making agents can assist humans in a large variety of applications: from replacing humans in dangerous missions to challenging them as game opponents or helping them make the best decision in complex situations.

In this context, learning algorithms have demonstrated promising results by allowing robots or agents to accumulate knowledge and learn how to perform effectively in a wide range of situations. However, most learning algorithms focus on learning one optimal solution to a problem: for example, one optimal strategy to solve a game, or one optimal gait to walk with a robot.

In this talk, we argue that instead, finding a diversity of solutions to the same problem is key to the development of learning algorithms across applications. Instead of finding one optimal solution to a problem, learning algorithms should seek to find thousands of diverse solutions, each leading to a different behaviour: for example, thousands of strategies for a single game, or thousands of different gaits to walk with a single robot.

We give multiple arguments for this diversity: interestingness, exploration, adaptation to out-of-distribution events, and downstream re-utilisation of learned behaviour for complex tasks. We also provide some pointers toward methods that encourage and maintain diversity in learning algorithms for robots and decision-making agents.

Shreya Iyler

PhD Student at University of Nottingham

Abstract: Machine Learning and Feature Importance Techniques for Variable Selection in Bullous Pemphigoid Analysis

Bullous Pemphigoid (BP) is a chronic autoimmune skin disorder, necessitating advanced analytical techniques to discern underlying patterns and potential predictive factors.

This study employs supervised machine learning classifiers, including Random Forests and logistic regression, alongside innovative post-training interpretation methods, such as SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME), facilitates the identification of key features impacting BP positivity.

The 30 key features were then subjected to cluster analysis to identify sub-clusters in the BP population. Furthermore, cluster analysis methods, including K-Means, DBSCAN, and HDBSCAN, are applied to segment the dataset, revealing distinct groupings and underlying structures.

Dimensionality reduction techniques like PCA and t-SNE aid in visualizing clustered data in lower dimensions.

The investigation uncovers six distinct clusters, each characterized by unique attributes and correlations, shedding light on potential outliers and noteworthy medication profiles.

Notably, the study underscores the significance of pivotal variables like “Penicillinase” and “Statins” in shaping cluster formations among BP patients, emphasizing the differential reliance on medications and vaccines for classification between BP populations and controls.

The findings suggest the DBSCAN method combined with PCA in three dimensions as the most effective approach for clustering.

Cara Lynch

PhD Student at UCL Computer Science

Abstract: Unravelling the Threads: Exploring Parallels Between Knitting and Modelling Online Social Networks

Online Social Networks are omnipresent in contemporary society, forming digital spaces where people gather to share information, interests and opinions.

As of January 2024, more than 84% of all adults use social media globally, and the typical user spends over two hours a day on these platforms.

Their impact extends far beyond the digital medium, with real-world effects observed in individuals, populations and economies.

Conversely, hand knitting can be seen as an antithesis to social media. A slow, repetitive, and distinctly tangible process, it offers an opportunity for creation detached from the digital realm.

Whether it’s unravelling a doomed tension swatch or the complexities of thread popularity, this talk aims to intertwine these seemingly disparate realms.

We'll delve into the unexpected connections between learning to knit and modelling post success on Reddit, a popular social media platform, illustrating the value of trial and error along the way.

 

Professor Louise Brown

Assistant Professor at University of Nottingham, Composites Research Group, Nottingham, Society of Research Software Engineers, Secretary of BCS Women

Louise Brown has a mechanical engineering background, completing both bachelors and PhD degrees at Nottingham. After completion of her PhD, she worked for several years in the Computer Science department at Nottingham working on automated design tools for CAD systems for industrial embroidery machines.

Following the birth of her daughter, who had various health issues, she worked from home as a self-employed software engineer for thirteen years. Louise returned to academia in 2009 to a research role combining her software engineering and composite materials expertise, responsible for the development of open source software TexGen. 

Louise is a professional member of the BCS, The Chartered Institute for IT and has been a committee member of BCS Women for about 15 years. The RSE community has grown enormously in recent years and the network is great for finding help and advice.

Louise's talk will look at some of the software that she's worked on during her career, a surprising amount of it to do with textiles in one way or another – CAD systems for industrial embroidery machines, software for home crafts and now textile composites for the aerospace industry. Louise will talk about TexGen and how it has developed from a student project to software which now has a world-wide user base.