Category: Events

Student perspectives: Compass Annual Conference 2025

Posted on by samuel.perren

A post by Compass student Samuel Perren.

Introduction

We were delighted to welcome partners, colleagues and alumni to the Bristol Avon Gorge (Hotel du Vin) in October, for the Compass Conference 2025. This year’s event focused on ‘Data Science for Social Good’, a topic our students are deeply passionate about. The day provided a fantastic opportunity to consider the positive role data science, computational statistics and AI can play across social, cultural and industrial fields.

It was wonderful to see Compass partners and alumni returning to join us, representing a wide variety of supportive organisations from across the private and public sectors – and to reflect on a year filled with notable achievements for our students, including a high number of papers accepted by prestigious conferences such as ICLR, ICML and NeurIPS.

Compass CDT Director, Professor Nick Whiteley, opening the Conference proceedings.

 

Student Research Talks

The morning started with four research talks from Compass students, showcasing the breadth of work being undertaken in the CDT. Samuel Perren opened the session with his talk on “Network Meta-analysis with Class Effects“, proposing a systematic approach to model selection. Following this, Ollie Baker presented “Rates of Convergence for Entropy Limits of Soft Random Geometric Graphs”, discussing the effects of node distribution and underlying geometry.

Yuqi Zhang then spoke on “Multiscale Detection of Multiple Change Points in High-dimensional Factor Models”, demonstrating a method that recovers change points even in dependent, high-dimensional data. Finally, Kieran Morris explored “Benamou Brenier Optimal Transport on Discrete Domains”, looking at how fluid flow interpretations of Optimal Transport adapt to discrete spaces.

Yuqi Zhang presents her research to an audience of Compass partners, alumni, students and colleagues.

 

Student Lightning Talks

The lightning talks session challenged students to present their work in a concise and engaging format. The topics were diverse, highlighting the versatility of our cohorts. Emma Tarmey discussed “Confounder Handling in Real-world Epidemiology”, using the OpenSAFELY NHS data platform. Ben Anson posed critical questions about resources with his talk “Matmul at Scale”, asking if AI could have converged on operations other than matrix multiplication.

Sherman Khoo provided an intuitive overview of “Fun with CLIP Embeddings”, exploring their practical applications. We also heard from Rachel Wood on “How to Help Others Use Data Science for Social Good”, sharing insights from a summer project helping qualitative researchers use data science.

Sherman Khoo presents his lightning talk: “Fun with CLIP Embeddings”

 

Guest Speaker Presentations

We were privileged to host an exciting line-up of spotlight and guest speakers, who brought a range of expertise to the conference theme. Colm Connaughton, Director at the London Mathematical Laboratory, discussed his time co-organising the Data Science for Social Good summer fellowship programme. Kevin Parry, Director of Data at the Information Commissioner’s Office (ICO), discussed leading the ICO’s Enterprise Data Strategy and responsible innovation.

Later in the afternoon, Marc Sarazin,  Senior Data Scientist at the Office for National Statistics (ONS) spoke on modernising statistical production using Generative AI , followed by Kiran Dhanjal-Adams, Future Leader Fellow at Royal Botanic Gardens, Kew, who illustrated how deep learning tools can answer questions about biodiversity conservation.

Finally, Alex Bowring, Senior Mathematical Consultant at the Smith Institute, delivered a talk on managing economical, green and reliable energy grids, and the use of dynamic reserve models,  and Alice Kendal, Senior Data Scientist at ethical AI company Trilateral Research, presented case studies about her work in using predictive risk modelling to improve safeguarding, mapping organised crime, and taking a data driven approach to tackling exploitation and violence.

Alex Bowring from the Smith Institute, delivered an interesting talk on energy grids.

 

Discussion Panel

The day featured a robust discussion panel, chaired by Compass CDT Director, Professor Nick Whiteley, and featuring all six guest speakers, focused on the theme “Data Science for Social Good”. The discussion explored the influence of data science on the modern world, considering both successes and lessons learned. The panellists provided insights into how projects can deliberately set out to have a social good angle, and how policy and practice help shape the world in which we conduct data science.

Discussion panel (left to right): Colm Connaughton, Kevin Parry, Marc Sarazin, Alex Bowring, Kiran Dhanjal-Adams, Alice Kendal and Nick Whiteley (Chair).

 

Concluding Remarks

The conference concluded with remarks from Nick, who summarised the day’s talks and discussions. The event successfully highlighted the broad opportunities presented by the theme and stimulated further exploration into how AI and computational statistics can have a positive social influence. We would like to thank all the speakers, partners, alumni, and students who joined us to make the Compass Conference 2025 a success.

Past conferences

Student perspectives: AI UK 2025 Conference

Posted on by shaun.jordan

A post by Sam Bowyer, PhD student on the Compass programme.

Compass at AI UK

The Alan Turing Institute’s AI UK 2025 Conference was held last month in the QEII Centre, Westminster, and three Compass students – Emma Ceccherini, Sherman Khoo, and myself – were present for both days of the event. We attended a variety of sessions and spent time exploring the exhibition stalls, which showcased a wide range of AI projects from within academia, government and industry.

Compass students and staff pictured at the AI UK 2025 Conference
Compass CDT students and staff at the AI UK 2025 Conference. From left to right:
Dr Dan Lawson, Emma Ceccherini, Sam Bowyer, Sherman Khoo and Helen Mawdlsey

It was an eye-opening experience to learn about the work that The Alan Turing Institute does, and especially insightful to see the myriad downstream applications of the machine learning theory that we spend so much time thinking about.

Conference highlights

One particular favourite exhibition was that of the Ministry of Justice (MoJ). Emma and I talked to a data scientist at the MoJ who was working on a tool that uses LLMs to explain laws in plain English, in order to help regular people better understand their rights.

Another project involved aggregating various disconnected datasets from across government on the national and local level in order to research social factors that might lead to successful post-prison rehabilitation or equally to recidivism.

It was encouraging to see a variety of projects and organisations at the conference aiming to use AI for social and public good, with a significant amount focussed on the climate and green-tech.

Whilst Compass wasn’t presenting at AIUK, colleagues from the Informed AI Hub, the Interactive AI CDT, the AI For Collective Intelligence (AI4CI) Hub, and Jean Golding Institute were.

It was great to not only see the other projects that are going on in the University, but also to be able to network with colleagues who only work down the road from the Fry Building (e.g. sharing Bristol restaurant recommendations)!

On the first day of the conference, Professor Charlotte Deane, Executive Chair of the Engineering and Physical Sciences Research Council (EPSRC), gave an informative keynote talk on the state of scientific research in UK academia. It was surprising to learn about the overall size of EPSRC and the range of activities they engage in, particularly their keenness for investing in spin-outs. I found Professor Deane’s talk to be very encouraging and optimistic.

The second day of the conference focused on governmental uses of AI, particularly in medicine and in defence. Professor the Lord Darzi, who recently led the Independent Investigation of the NHS in England, gave an incredibly thoughtful talk on the opportunities for AI within the NHS.

He likened the current AI boom to the development of keyhole surgery in the second half of the 20th century, urging fast, nationwide deployment in order to improve health outcomes and equality throughout the UK.

Three talks on defence and national security similarly stressed the importance of fast uptake of AI tools and made clear the desire for public-private partnerships (including with academia) in order to make this happen. (The importance of cross-sector collaboration was consistently a strong theme at AIUK, although the absence of frontier AI labs did, in my opinion, betray a slight limit to this stated commitment).

Presentation karaoke

It wasn’t all so serious, however! The conference finished its first day with “Presentation Karaoke”, in which eight contestants competed to present unseen 5-minute long, 10-slide PowerPoints, each more bizarre than the last.

This fun, often slightly cringe-inducing, activity is now rumoured to be deployed at a future COMPASS student event. (Get practising your stand-up now…)

In summary, AIUK was a great opportunity to see how AI/ML research leads to real-world impact in the UK, and I would recommend attending to any CDT student in the future.

Guest Lecture: Professor Chris Breward

Posted on by shaun.jordan

An Introduction to Knowledge Exchange

The Compass CDT was delighted to recently host a Guest Lecture by Professor Chris Breward, from the Mathematical Institute, University of Oxford.

Chris led an interactive session for our PhD students, which focused on getting started with knowledge exchange (KE), and explored the skills needed to engage with industrial and other external partners.

As Scientific Director of the Knowledge Exchange Hub for Mathematical Sciences and Co-Director of the EPSRC CDT in Industrially Focused Mathematical Modelling, Chris had a wide range of valuable advice to share.

Drawing on his experience building parternships with companies and setting-up projects with industry co-funding, he ran through the different ways researchers at all stages of their career can get involved in KE.

Attendees explored why companies might engage with mathematical scientists, discussed things to consider before meeting potential collaborators, and looked at what can sometimes go wrong with academic-business relationships.

Student reflections

“During his Guest Lecture, Chris chatted with all of us about ways to communicate with non-academics during shared projects and how to do positive work as mathematical consultants.

“The session covered the pragmatics and hard-skills of private sector contract work, as well as the soft skills of open body language, effective listening and people management.

“He described the barriers that can arise between researchers (mathematicians in particular) and industrial partners. We then chatted interactively through where these pitfalls come from and how best to avoid them.

“He also gave us an entry-level look into the broader differences between universities and industry.”

KE initiatives

Chris closed by encouraging attendees to get involved with some of the opportunities the KE Hub provides for PhD students and researchers, such as the online Triage Workshops. These events can provide a safe space for individuals to gain experience with knowledge exchange, by observing senior colleagues from across the country.

He expressed his hope that Compass students would benefit from the upcoming five-day European Study Group with Industry (ESGI), which will take place here at the University of Bristol from Saturday, 14 July to Wednesday, 18 July 2025.

The Compass CDT was grateful to Chris for giving up his time to visit us in the School of Mathematics’ Fry Building, and we look forward to seeing him in Bristol again in the future.

As well as being an applied mathematician, lecturer and researcher at University of Oxford, Chris is co-founding Chief Moderator of the Mathematics-In-Industry Reports online KE repository, and a member of the Newton Gateway’s Scientific Advisory Board.

Student perspectives: Compass Annual Conference 2024

Posted on by ben.anson

A post by Compass students Ben Anson, Ollie Baker, Codie Wood and Rachel Wood.

Introduction

This October, we held our third annual Compass Conference. Unlike previous years, when the event was held in the University’s Fry Building, this time it took place at M Shed, offering scenic views of Bristol harbour. It was a great day for past and present Compass students, academics, and industry partners to come together and discuss this year’s theme: “The Future of Data Science”. With recent advances in machine learning and AI, it felt like a fitting time to learn from each other’s perspectives and to share ideas about how to move forward in this exciting space.

Panoramic view of Bristol harbour, as seen from M Shed

Student Research Talks

The morning started with four ten-minute research talks from Compass students. First was Rahil Morjaria‘s talk on “Group Testing” which explored current developments in the field, including algorithms and information-theoretic limits.

Following this, Kieran Morris presented “A Trip to Bregman Geometry and Applications”, considering advancements such as natural gradient methods, Bregman K-means clustering, and EM-projection algorithms that Bregman Geometry has enabled.

Ettore Fincato talked us through “Gradient-Free Optimisation via Integration”, focusing on a novel yet easy-to-implement algorithm for optimisation using Monte Carlo methods. Finally, Ed Milsom spoke about “Data Modalities and the Bias-Variance Decomposition”, taking us through a history of neural networks and speculating about why certain data types are so powerful, and why the future of general-purpose AI must be multi-modal.

Student Lightning Talks

The lightning talks challenged ten students to present on a topic in just three minutes. The ability to quickly convey a message in an engaging and understandable manner, to an audience with diverse backgrounds, is crucial in both academia and industry, and the students rose to the occasion.

Their talks captured the interest of the audience and inspired interesting questions that forced the students to think on their feet. Topics ranged from neural networks and large language models (LLMs), to making music using mathematics.

Compass Alumni Panel

This year’s conference panel, chaired by Compass CDT Director, Professor Nick Whiteley, offered an engaging look into the professional journeys of Compass alumni Dominic Owens, Jake Spiteri and Michael Whitehouse since  completing their PhDs. With shared experiences in finance, each panelist provided unique insights into the early career landscape and the skills that helped them succeed.

Jake delved into the details of his day-to-day work in the financial sector, while Dominic discussed the challenge and dedication required to secure a role through extensive networking and job applications. Michael shared details of his transition from finance to epidemiological research. Together, they sparked valuable discussions on what the future of data science might hold for upcoming Compass graduates.

Special Guest Lecture

The conference concluded with an enlightening special guest lecture by Professor Aline Villavicencio, Director of the Institute for Data Science and Artificial Intelligence at the University of Exeter. Her talk, “Testing the Idiomatic Language Limits of Foundation Models: The Strange Case of the Idiomatic Eager Beaver in Cloud Nine,” offered a fascinating counterpoint to the current enthusiasm surrounding LLMs.

Drawing from her research in Natural Language Processing (NLP), Professor Villavicencio demonstrated how even today’s most advanced models struggle with aspects of language that humans master naturally – particularly idioms and multi-word expressions. She illustrated a persistent gap between machine and human linguistic capabilities, reminding us that the path to truly human-like language understanding remains long and complex.

She also shared her perspective on the cyclical nature of NLP research, noting how, throughout her career, there have been multiple predictions about NLP research becoming obsolete as models improve. Yet, as her work on datasets like SemEval (Semantic Evaluation) shows, there remain fundamental challenges in representing and understanding idiomatic language.

Concluding remarks

The successful day of talks, poster sessions and networking culminated with Professor Whiteley sharing his thoughts on what we learned throughout the event. He concluded that the future of our field is certain to be exciting and will encompass a huge range of different areas and ideas. This year’s conference embodied this by providing a platform for students, academics, and industry professionals to share new insights from many different sectors, and to form strong relationships to help forge a path to the future of data science.

 

Past conferences

Compass Away Day 2024

Posted on by sam.bowyer

A post by Sam Bowyer and Emma Ceccherini, PhD students on the Compass programme.

The annual Compass Away Day took place this past June at Folly Farm, a fully sustainable and eco-friendly venue that offered Compass students the opportunity to take some time away from their regular research and enjoy a variety of activities in the Somerset countryside. Over the course of three days, the students learnt—among other things—how to craft an effective CV as a machine learning researcher; how each of our research areas overlap in surprising ways; how to improve the execution time of Python programs; and how to throw an axe.

On arrival, the students first took part in a Responsible Innovation talk led by Henry Bourne, in which we explored the connections, biases, and gaps in and between our individual research areas. This took the form of a creative mind-mapping exercise in which the students eagerly relished the opportunity to don party hats (see picture below).

After a hearty lunch with plenty of coffee, the students next heard three short talks by their colleagues. First, Ed Davis treated them to a talk on the benefits of just-in-time compilation, with a live demonstration of how one line of code (‘@jit‘) can speed up your Python scripts by multiple orders of magnitude. Secondly, the students saw another live-coding demonstration given by Kieran Morris as he walked through a project of his which utilised command-line automation and LLM APIs to create a text-based simulation of a well-known sporting event (The Hunger Games, but with Compass PhD students appearing as competitors). Finally, Edward Milsom gave a tongue-in-cheek TED-style talk entitled ‘How to be an AI Bro’ advising all present in the audience to pick a side between Doomers and Zoomers (anti- and pro-AI); to use self-attention mechanisms (the basis of modern LLMs; see https://arxiv.org/abs/1706.03762) whenever possible; and, following that last piece of advice, to follow @edward_milsom on X (formerly known as Twitter).

The final session of the day was led by Emma Ceccherini and Sam Bowyer, in which the two Away Day student organisers presented on the topic of CVs and online presence. The presentation exceeded everyone’s expectations, in no small part thanks to Helen Mawdsley being present to answer questions from the audience.

Feeling refreshed and revitalised after a peaceful night’s sleep, the students awoke to sunshine and birdsong. But soon a competitive feeling came over the group; this next morning was devoted to sports: archery, axe-throwing and segwaying. We won’t dwell on the sports too much except to say that Sam Bowyer’s team won handily, and that one impressed student is reported to have said “[Sam] was a G at archery. Bosh.” (That same anonymous source, taking time out of her quantitative spatial science and networks research—located in the big Compass office somewhere between desks C and E—, is also reported to have said “Ed D[redacted for anonymity], snapping all those single-use poles, not impressive segwaying.” We, the editors, feel it would be improper for us to comment on this matter.)

The final activity on the schedule was a writing retreat, taking place over the afternoon of the second day and the entirety of the third and final day. Being in a peaceful location and setting time aside to work on a single specific task proved useful to the students present.

To summarise, Away Day 2024 went extremely well, having been organised expertly by Crina Radu, to whom the students are all incredibly grateful.

Compass is funded by EPSRC
Skip to toolbar