IEEE UK & Ireland Signal Processing Society Early Career Research Seminar
The IEEE UK&I Signal Processing Society Chapter is pleased to launch a series of online Early Career Research (ECR) seminars, offering early career researchers, including PhD students, post-doctoral researchers, Fellows, and new lecturers, a platform to present technical talks to the broader signal processing community.
As part of our ongoing commitment to diversity, equity, and inclusion, the IEEE UK&I SPS Chapter is proud to highlight and promote the contributions of women in signal processing. This initiative not only provides a platform for emerging talent but also aims to inspire and empower the next generation of women engineers and researchers in the field. By showcasing diverse voices and perspectives, we hope to offer a more inclusive and vibrant research community.
Therefore, as part of this initiative, kicking off the series, we are delighted to welcome the Dr. Shuyan Li and Dr. Xinming Shi from Queen’s University Belfast, presenting on “Training-Free Event-Aware Video Anomaly Detection” and “Evolutionary Brain-Inspired Computing Systems Based on Memristors”, respectively.
We hope these events will contribute to refreshing and strengthening the UK signal processing research landscape and look forward to welcoming you.
Speaker 1: Dr. Shuyan Li
Title: Training-Free Event-Aware Video Anomaly Detection
Abstract: Video Anomaly Detection (VAD) focuses on identifying anomalies within videos. Supervised methods require an amount of in-domain training data and often struggle to generalize to unseen anomalies. In contrast, training-free methods leverage the intrinsic world knowledge of large language models (LLMs) to detect anomalies but face challenges in localizing fine-grained visual transitions and diverse events. Therefore, we propose EventVAD, an event-aware video anomaly detection framework that combines tailored dynamic graph architectures and multimodal LLMs through temporal-event reasoning. Specifically, EventVAD first employs dynamic spatiotemporal graph modeling with time-decay constraints to capture event-aware video features. Then, it performs adaptive noise filtering and uses signal ratio thresholding to detect event boundaries via unsupervised statistical features. The statistical boundary detection module reduces the complexity of processing long videos for MLLMs and improves their temporal reasoning through event consistency. Finally, it utilizes a hierarchical prompting strategy to guide MLLMs in performing reasoning before determining final decisions. We conducted extensive experiments on the UCF-Crime and XD-Violence datasets. The results demonstrate that EventVAD with a 7B MLLM achieves state-of-the-art (SOTA) in training-free settings, outperforming strong baselines that use 7B or larger MLLMs.
Speaker 2: Dr. Xinming Shi
Title: Evolutionary Brain-Inspired Computing Systems Based on Memristors
Abstract: As Moore's Law approaches its physical limits, traditional computing architectures encounter significant bottlenecks, particularly the von Neumann bottleneck in data-intensive applications. My research introduces an evolutionary brain-inspired computing system utilizing memristors to overcome these challenges. This innovative approach draws inspiration from both the brain and nature to forge an adaptive and efficient architecture, incorporating basic units, functional modules, and a next-generation spiking neural network. By integrating principles of evolutionary computing, this system enhances its reliability, explainability, and robustness. This work aims to revolutionize intelligent computing, ushering in a new era of trustworthy and efficient solutions.
Date and Time
Location
Hosts
Registration
-
Add Event to Calendar
Speakers
Dr. Shuyan Li of Queen's University Belfast
Training-Free Event-Aware Video Anomaly Detection
Video Anomaly Detection (VAD) focuses on identifying anomalies within videos. Supervised methods require an amount of in-domain training data and often struggle to generalize to unseen anomalies. In contrast, training-free methods leverage the intrinsic world knowledge of large language models (LLMs) to detect anomalies but face challenges in localizing fine-grained visual transitions and diverse events. Therefore, we propose EventVAD, an event-aware video anomaly detection framework that combines tailored dynamic graph architectures and multimodal LLMs through temporal-event reasoning. Specifically, EventVAD first employs dynamic spatiotemporal graph modeling with time-decay constraints to capture event-aware video features. Then, it performs adaptive noise filtering and uses signal ratio thresholding to detect event boundaries via unsupervised statistical features. The statistical boundary detection module reduces the complexity of processing long videos for MLLMs and improves their temporal reasoning through event consistency. Finally, it utilizes a hierarchical prompting strategy to guide MLLMs in performing reasoning before determining final decisions. We conducted extensive experiments on the UCF-Crime and XD-Violence datasets. The results demonstrate that EventVAD with a 7B MLLM achieves state-of-the-art (SOTA) in training-free settings, outperforming strong baselines that use 7B or larger MLLMs.
Biography:
Shuyan Li is a Lecturer in the School of Electronics, Electrical Engineering and Computer Science (EEECS) at Queen’s University Belfast. Prior to this, she worked as a Postdoctoral Research Associate in the Department of Engineering at the University of Cambridge, collaborating with Prof. Ioannis Brilakis. In Jan. 2022, she got her doctor’s degree in the Department of Automation at Tsinghua University, Intelligent Vision Group(IVG). She was supervised by Prof. Jiwen Lu , Prof. Jie Zhou and Prof. Xiu Li . Shuyan is broadly interested in computer vision and AI for science. Her current research focuses on representation learning, video understanding, medical imaging and construction digital twins.
Email:
Address:United Kingdom
Dr. Xinming Shi of Queen's University Belfast
Evolutionary Brain-Inspired Computing Systems Based on Memristors
As Moore's Law approaches its physical limits, traditional computing architectures encounter significant bottlenecks, particularly the von Neumann bottleneck in data-intensive applications. My research introduces an evolutionary brain-inspired computing system utilizing memristors to overcome these challenges. This innovative approach draws inspiration from both the brain and nature to forge an adaptive and efficient architecture, incorporating basic units, functional modules, and a next-generation spiking neural network. By integrating principles of evolutionary computing, this system enhances its reliability, explainability, and robustness. This work aims to revolutionize intelligent computing, ushering in a new era of trustworthy and efficient solutions.
Biography:
Dr. Xinming Shi is an Assistant Professor at Queen's University Belfast. She earned her Ph.D. in Computer Science from the University of Birmingham, UK, in 2023. Her research is dedicated to brain-inspired intelligence, including neuromorphic computing hardware and software, evolutionary learning, and trustworthy brain-inspired systems. She is a 2024 recipient of the Leverhulme Early Career Fellowship from the Leverhulme Trust. Dr. Shi has published extensively in top journals and conferences, including IEEE TNNLS, IEEE TC, IEEE TETCI, and ACM Transactions. She is a member of SIGEVO, part of the Conference Activities and Communications Subcommittee of the IEEE Computational Intelligence Society (CIS), and a Youth Editor of Intelligent Control. Additionally, she has served on various committees and as a reviewer for leading journals and conferences.
Email:
Address:United Kingdom
Please register for the event using the following Registration Link