Inter-Group Presentation Workshop

The Inter-Group Presentation Workshop continues to bring together PhD students from different EECS research groups, providing a friendly and supportive environment to share research, exchange ideas, receive constructive feedback, and learn about the exciting work taking place across the School. The workshop encourages interdisciplinary discussion, provides opportunities to receive constructive feedback from peers, and helps students build confidence in communicating their research to diverse audiences.

The workshop also offers a safe space to practise articulating and defending research in preparation for key academic milestones, including conferences, progression reviews, and viva examinations. By connecting students from different research areas, it promotes collaboration, broadens research perspectives, and strengthens the EECS postgraduate community.

Please get in touch with Sofia or Dario if you would like to present your current research. It is a great opportunity to rehearse presentations, receive feedback in a supportive environment, and showcase your research to colleagues from across EECS.

Next Session

Our next speaker, in collaboration with the CogSci group, is Hugo Enrique Ramirez- Centeno 🎉, a final-year PhD student in the Networks group at QMUL. Hugo will be presenting his research on Argument Minecrafting in the Internet Governance ecosystem.

• Title of talk: Minecrafting arguments in the Internet Governance Ecosystem

• When: Wednesday 4th November 2026, 13:00-14:00.

• Where: PL 4.24, Peter Landin Building.

Speaker: Hugo Enrique Ramirez- Centeno (Link)

Bio: Hugo is supervised by Dr Ignacio Castro. His research focuses on Natural Language Processing (NLP), dialogue modeling, email analytics, and conversational dynamics in technical and internet governance communities (such as the IETF and ICANN). Outside of research, he enjoys travelling to historical places in the city and visiting museums and gardens to learn the history of the place through its architecture. He also likes video games and music as well as the London Nightlife.

Abstract: The Internet Governance communities aim to gather consensus among all participants to make the internet work, but are the publications and decisions really driven by all members of the community or by a small group?. This talk aims to explore some techniques to measure the email interactions of the IETF, we aim to characterize interactions within countries and affiliations to measure any differences between the parties. 

Previous Session

The twelfth session of our Workshop was a Great Success 🎉! 

17 participants attended the session: 11 in person and 6 remotely 🌟.

A special thanks to our speaker Roksana Goworek🎉, a final-year PhD student in the centre for Human-Centered computing (HCC) at QMUL. Roksana is currently interning at the Alan Turing Institute on mechanistic interpretability. She presented her work on measuring and learning lexical meaning across time and languages.

Speaker: Roksana Goworek (Link)

• Title of talk: Meaning What, Exactly? Measuring and Learning Lexical Meaning Across Time and Languages

• When: Wednesday 21st October 2026, 13:00-14:00.

• Where: PL 4.24, Peter Landin Building.

Bio: Roksana is a final-year PhD student at QMUL, supervised by Dr. Haim Dubossarsky, working on word meaning in large language models, including creating and evaluating methods for analysing corpora differences, investigating why certain abilities transfer across some languages more than others, and recently analysing how contextual information is integrated into representations.

 Abstract: How can we analyse changes in word meaning across diachronic corpora? Contextualised language models have enabled large-scale approaches to lexical semantic change detection, but turning their rich word representations into reliable evidence of change raises several methodological questions. How should sets of representations be compared, how can models capture sense distinctions when target-language data are scarce, and how can we efficiently construct the high-quality annotated data needed for training and evaluation? This talk explores how choices in measurement, model training, and data annotation shape our ability to detect semantic change, particularly in low-resource languages, and what is required to extend such research reliably to a wider range of languages.

The eleventh session of our Workshop was a Great Success 🎉! 

17 participants attended the session: 10 in person and 7 remotely 🌟.

Special thanks to our speaker, @Siyou Li 🎉, a second-year PhD student in the Cognitive Science Research Group at Queen Mary University of London. He presented his accepted paper at the MICCAI 2025 conference in Daejeon, South Korea. His paper proposes a differentiable multi-scale multi-modal tokenizer (μ²) for automated radiology report generation. The work further uses Direct Preference Optimization (DPO), guided by the GREEN medical report evaluation metric, to improve the clinical quality of generated radiology reports.

Speaker: Siyou Li (Link)

• Title of talk: μ²Tokenizer: Differentiable Multi-Scale Multi-Modal Tokenizer for Radiology Report Generation

• When: Tuesday, 6th October 2026, 13:00–14:00

• Where: PL 4.24, Peter Landin Building

Bio: Siyou Li is a second-year PhD student in the Cognitive Science Research Group at Queen Mary University of London. His research focuses on multimodal large language models and medical image understanding, particularly on connecting high-dimensional 3D medical imaging with language. His recent research explores efficient visual tokenization, multimodal representation learning, and preference optimization for automated radiology report generation.

The tenth session of our Workshop was a Great Success 🎉! 

19 participants attended the session: 14 in person and 5 remotely 🌟.

Special thanks to our speaker, @Zhansaya Amangeldiyeva 🎉, also known as 'Saya', a PhD student in a joint doctoral programme between the University of Genoa, Italy, and Queen Mary University of London (QMUL). She presented her recently submitted paper, under review at the GlobeCom 2026 conference in Macau S.A.R., China, which presents a Proximal Policy Optimisation (PPO) reinforcement learning framework that repositions a mobile gateway in real time to maximise TDoA localisation accuracy.

Speaker: Zhansaya Amangeldiyeva (Link).

• Title of talk: Self-Discovering Mobile Gateway Positioning for LoRaWAN TDoA Localization Using PPO

• When: Tuesday 21st   July 2026, 13:00-14:00.

• Where: PL 4.24, Peter Landin Building.

Bio: Zhansaya Amangeldiyeva is a PhD student on a joint doctoral programme between the University of Genoa and QMUL, working on wireless communication and reinforcement learning. Her research interests include reinforcement learning for UAV communications, LoRaWAN localisation, and network security and anomaly detection for IoT systems. Her previous research focused on cybersecurity and cyberattacks on IoT devices. She is also a Research Assistant at QMUL. She was a member of the FollowMe project. The FollowMe project addresses these security issues by providing methods and tools to prevent crimes on LoRaWAN networks, as well as using drone technology to identify and track attackers. Outside her academic work, she enjoys organising events as an active member of the Walk and Talk Committee, hiking, swimming, exercising, and spending time with friends and colleagues.

The ninth session of our Workshop was a Great Success 🎉!

 18 participants attended the session: 11 in person and 7 remotely 🌟. 

A special thank you to the speaker, @Debin Meng🎉, a final-year PhD candidate in the MMV group and a research assistant at Queen Mary University of London. Debin presented his recently accepted paper at the IJCNN 2026 conference in the Netherlands, which proposes a training-free approach that improves generation diversity while maintaining competitive image quality, leading to a better diversity–quality trade-off.

Speaker: Debin Meng (Link).

• Title of talk: TPSO: Training-Free Diverse Image Generation via Semantic Prompt Embedding Optimization.

• When: Monday 20th July 2026, 13:00-14:00.

• Where: PL 4.24, Peter Landin Building.

Bio: Debin Meng is a final-year PhD candidate and research assistant at Queen Mary University of London, working on computer vision and generative AI. His current research focuses on image generation, including text-to-image generation, multimodal image generation and editing, generative model evaluation, and diversity-aware generation. His previous research focused on understanding facial behaviour. He is also a Research Assistant on an EU Horizon 2020 project, working on efficient and lifelong learning methods. Outside his academic work, he enjoys hiking, swimming, exercising, and spending time with friends and colleagues.

The seventh session of our Workshop was a Great Success 🎉! 

19 participants attended the session: 13 in person and 6 remotely 🌟. 

A special thank you to @Wenqing Wu, who delivered an engaging presentation on presenting his paper: NovBench: Evaluating Large Language Models on Academic Paper Novelty Assessment, an accepted paper at the Findings of ACL 2026 conference in California, USA. The talk explored the evaluation of large language models for assessing the novelty of academic research papers. 

Speaker: Wenqing Wu (Link).

• Title of talk: NovBench: Evaluating Large Language Models on Academic Paper Novelty Assessment.

• When: Tuesday 23rd June 2026, 13:00-14:00.

• Where: PL 4.24, Peter Landin Building.

Bio: Wenqing Wu is a final-year PhD student at the School of Economics and Management, at NJUST, China. He is currently an academic visitor at EECS, Queen Mary University of London. His research interests include natural language processing, academic paper novelty assessment, peer-review text mining, and large language model evaluation. Outside of research, he enjoys playing basketball and working out at the gym.

The sixth session of our Workshop was a Great Success 🎉! 

20 participants attended the session: 17 in person and 3 remotely🌟. 

A special thank you to @Yaqi Sun for delivering an engaging presentation on Meta-Learning for Multi-Agent Reinforcement Learning. The talk explored how multi-agent systems can learn to adapt across different tasks and environments, generating valuable discussion among attendees on the future of adaptive and intelligent learning systems.

Speaker: Yaqi Sun (Link).

@Yaqi Sun 🎉, is a PhD student in the Machine Intelligence and Decision Systems (MInDS) group, working on meta-learning for multi-agent reinforcement learning. She presented her PhD research on how multi-agent systems can learn to adapt their learning strategies across different tasks and environments.

• Title of talk: Meta-Learning for Multi-Agent Reinforcement Learning.

• When: Tuesday 16th June 2026, 13:00-14:00.

• Where: PL 4.24, Peter Landin Building.

Bio: Yaqi Sun is a PhD student at Queen Mary University of London. Her research focuses on meta-learning and multi-agent reinforcement learning, with a particular interest in developing adaptive learning methods that improve the efficiency, robustness, and generalisation of multi-agent systems. Outside of research, she enjoys coffee, hiking, playing badminton and tennis.

The fifth session of our Workshop was a Great Success 🎉! 

16 participants attended the session: 13 in person and 3 remotely🌟. 

A special thank you to @Carlos Dazaeth Quintal-Gueta, who delivered an engaging presentation on AI-driven lesion detection in medical images, providing an accessible introduction to modern object detection techniques, the medical image analysis pipeline, and the use of CNN- and Transformer-based models for identifying lesions in clinical imaging data.

Speaker: Carlos Dazaeth Quintal (LinkedIn).

@Carlos Dazaeth Quintal Gueta🎉, a Research Master's (ResM) student in the MMV group undertaking a dual programme between the Instituto Politécnico Nacional (IPN), Mexico and QMUL. Daz will be presenting his current research on AI for lesion detection in medical images.

• Title of talk: AI-driven lesion detection.

• When: Tuesday 2nd June 2026, 13:00-14:00.

• Where: PL 4.24, Peter Landin Building.

Bio: Carlos is a Visiting Research student from IPN, working on computer vision and deep learning, with a particular interest in medical applications of AI. His research focuses on developing intelligent systems for medical image analysis and lesion detection. Outside of research, he enjoys listening to music and hiking.

The fourth session of our Workshop was a Great Success 🎉!  

We had 16 participants in attendance; 12 in person and 4 remotely🌟. 

Speaker: Daniel Gill (Website).

A special thank you to @Daniel Allan Gill, who delivered an engaging presentation on the Identification of Differences in Haptic Interaction for Neurodivergent Adults. The talk explored how autistic and other neurodivergent individuals may experience sensory and motor interactions differently, generating valuable discussion among attendees.

Daniel Allan Gill, is a second-year PhD student, the chief representative of EECS PhD students, and an active ex-member of the walk and talk committee. Dan presented his current research project to EECS PhD students. 

• Title of talk: Identification of Differences in Haptic Interaction for Neurodivergent Adults.

• When: Tuesday 19th May 2026, 13:00-14:00.

• Where: PL 4.24, Peter Landin Building.

Bio: Daniel is a PhD candidate and researcher at Queen Mary University of London, working on robotics and psychology. He aims to understand where neurodivergent people have different sensory and motor experiences experimentally using robotic systems, to design and develop inclusive and truly useful technologies. Daniel also co-runs the QMUL MINDS network, building on his own experiences of a disconnect between neurodivergent people and researchers. Through MINDS, he has hosted inclusive workshops where these two groups can discuss research in an equal space. He’s also done (surprisingly okay) autism/science-themed stand-up comedy on a few occasions. When not working, Dan enjoys gardening and exploring the nature.

Past Sessions:

The third session of our Workshop was a Great Success 🎉!  

We had 19 participants in attendance; 14 in person and 5 remotely🌟. 

Yuli Sutoto Nugroho, gave a great presentation on his current research concerning the importance of having lecturer and classmates while learning.

Speaker: Yuli Sutoto Nugroho

Title of talk: Is it Important to Have a Lecturer and Classmates in Your Class?  

When: Tuesday 5th May 2026, 13:00-14:00 

Location: PL 4.24, Peter Landin Building 

Bio: 

Yuli Sutoto Nugroho is a PhD student at Queen Mary University of London’s School of Electronic Engineering and Computer Science. His research focuses on educational technology, distance learning, vocational education, eye-tracking, and facial expression recognition. His PhD thesis explores the impact of presence in video-based learning and the metaverse. Alongside his studies, he serves as a lecturer at the State University of Surabaya, Indonesia. He is also a content creator on his social media.  

The second session of our Inter-Group Presentation Workshop was a great success 🎉! We had 15 participants in attendance, (9 in person and 6 remotely). 

Yufeng-Li, from the Cogsci group working with @Arkaitz, presented her recently accepted paper at ICASSP with confidence and a clear, detailed explanation. 

Speaker: Yufeng Li

Title of the paper : RAVE: Retrieval and Scoring Aware Verifiable Claim Detection

When: Tuesday 21st April 2026, 11:00-12:00.

Location: PL 3.02, Peter Landin Building.

Bio:

Yufeng Li is a third-year PhD student working on natural language processing, with a particular interest in automated fact-checking and claim detection. When not working, she enjoys bouldering and rope climbing.

Abstract:

The rapid spread of misinformation on social media underscores the need for scalable fact-checking tools. A key step is claim detection, which identifies statements that can be objectively verified. Prior approaches often rely on linguistic cues or claim check-worthiness, but these struggle with vague political discourse and diverse formats such as tweets. We present RAVE (Retrieval and Scoring Aware Verifiable Claim Detection), a framework that combines evidence retrieval with structured signals of relevance and source credibility. Experiments on CT22-test and PoliClaim-test show that RAVE consistently outperforms text-only and retrieval-based baselines in both accuracy and F1.

The session opened with great speech from Dr Raymond Hu, who welcomed the initiative and highlighted its importance within EECS. He emphasised that the workshop provides a valuable opportunity for PhD students in EECS to share research ideas across groups, engage in discussions, and receive peer feedback. He further noted that such initiatives are essential for building confidence in academic communication, particularly in preparation for conferences, milestones, and the final viva. The workshop was positioned as a community-driven platform to encourage collaboration and interaction among EECS PhD students.

Speaker: Dr. Richard Clegg

Title of talk: How to do presentations?

When: Tuesday 14th April 2026, 13:00-14:00.

Location: PL 4.24, Peter Landin Building.

Bio:

Richard Clegg is a Senior Lecturer in Networks at EECS. He teaches on the joint programme with Beijing. He works on problems in temporal networks, including social networks, financial networks, and computer networks, for example, detecting fraud in cryptocurrency networks. He is the research lead at the spin-out company Pometry. When not working he enjoys scuba diving and makes videos for the YouTube channel computerphile.