Recent papers, talks, awards, and activities from the group.
Congratulations to our PhD student Jiawen Qi, first author of “DeltaLLM: Delta-Key Sparse Attention for Edge LLMs,” on the paper’s acceptance for publication in the IEEE Transactions on Circuits and Systems for Artificial Intelligence!
Authors: Jiawen Qi, Jiayu Liu, Chang Gao, Zhaochun Ren, and Qinyu Chen.
Qinyu gave an invited talk titled “Low-latency LLMs at the Edge” at the IEEE CIS “Foundation Models Meet Robotics” summer school in Paderborn, Germany.
Congratulations to our MSc student Pu Li and his day supervisor, PhD student Jiawen Qi, on the acceptance of “When NPUs Are Not Always Faster: A Stage-Level Analysis of Mobile LLM Inference.”
We are also pleased to share that “MonarchGRU: A Low-Footprint Recurrent Neural Network for Audio Keyword Spotting” has been accepted. This work is a collaboration with Aalto University, Yonsei University, and TU Delft.
Qinyu gave a talk at NEAT Symposium 2026 in Utrecht. Her talk, “Interactive Edge Intelligence for Next-Generation Human–Machine Systems,” explored how to develop human-centered edge AI systems.
Optical Processing for Energy-Efficient Real-time Applications (OPERA)
Together with Prof. Dr. Allard P. Mosk, Utrecht University; Prof. Dr. Ryoichi Horisaki, The University of Tokyo; Prof. Dr. K. J. Batenburg, Leiden University; and Dr. Y. Shoji, Institute of Science Tokyo.
We recently had the pleasure of hosting Dr. Shintaro Shiba from The University of Tokyo and Woven by Toyota for an invited talk.
His talk, “Unlocking the Potential of Event Cameras: From Events to Motion to 3D Understanding,” highlighted recent advances in event-based vision and their applications in motion analysis and 3D understanding.
Our work, “SHAP-AAD: DeepSHAP-Guided Channel Reduction for EEG Auditory Attention Detection,” has been accepted for ISCAS 2026, Shanghai. Congratulations to our former BSc student Rayan Salmi and PhD student Guorui Lu.
A big thank you to Zhen, Jiawen, and Guorui for their voluntary work.
Qinyu gave an invited talk entitled “Efficient Interactive Intelligence at the Edge” at the School of Integrated Circuits, Tsinghua University.
Zhen Xu and Guorui Lu’s work on EvHand-FPV won the best poster award from the Edge AI for Robotics Workshop at IROS 2025 in Hangzhou, China.
The IEEE Journal on Emerging and Selected Topics in Circuits and Systems (JETCAS) invites submissions for the Circuits and Systems for Extended Reality Special Issue. Deadline: 2 March 2026.
Qinyu is the corresponding guest editor together with Abhronil Sengupta, The Pennsylvania State University; Ning Qiao, SynSense; and Hoi-Jun Yoo, KAIST, Korea.
Qinyu received funding through the Dutch Research Council (NWO) Veni Talent Programme for research on embedded LLM system design.
At ISCAS 2025 in London, UK, PhD student Guorui gave an oral presentation on his work titled “SlimSeiz: Efficient Channel-Adaptive Seizure Prediction Using a Mamba-Enhanced Network.”
During the annual meeting of the IEEE Circuits and Systems Society Neural Systems and Applications Technical Committee (IEEE CASS NSA TC), Qinyu was honored to be elected secretary of the NSA TC.
Invited by Dr. Xinming Shi, Qinyu gave an invited talk at Queen’s University Belfast, UK, on neuromorphic vision and eye-tracking technologies for XR applications.
Together with Guangzhi Tang (Maastricht University), Ali Safa (Hamad Bin Khalifa University), Neelesh Kumar (Procter & Gamble), and Fabian Clemens Weigend (Harvard), we will organize the “Edge AI for Robotics: Emerging Technology and Application” workshop at IROS in Hangzhou, China, this October. Many thanks to AMD for the sponsorship.
Our tutorial “Neuromorphic AI-Enhanced Eye Tracking Technology: Algorithms, Circuits and Systems” was successfully presented at AICAS 2025 in Bordeaux, France. Together with Zhuo Zou (Fudan University) and Guohao Lan (TU Delft), we introduced how neuromorphic vision and AI can transform eye tracking for ultra-efficient wearable applications.
MSc student Zhen Xu presented “Working in Progress: Event-based 3D Hand Tracking for Extended Reality” at the Neuro-inspired Computational Elements (NICE) 2025 conference in Heidelberg, Germany.
Our work, “FACET: Fast and Accurate Event-Based Eye Tracking Using Ellipse Modeling for Extended Reality,” has been accepted for ICRA 2025 in Atlanta, US. This collaboration includes Junyuan Ding (Beihang University), Ziteng Wang (DVSense), Chang Gao (TU Delft), Min Liu (DVSense), and Qinyu Chen.
Congratulations to PhD student Guorui Lu: his paper “SlimSeiz: Efficient Channel-Adaptive Seizure Prediction Using a Mamba-Enhanced Network,” with TU Delft and Shanghai Renji Hospital, has been accepted for a lecture presentation at ISCAS 2025 in London, UK.
Together with TU Delft, the University of Würzburg, DVSense, and Prophesee, we are organizing the Efficient Event-based Eye-Tracking (3ET) Challenge as part of the CVPR 2025 Event-based Vision Workshop. The challenge will run from mid-February to mid-March 2025. Stay tuned for more details and join us in advancing event-based vision technology.
The work titled “Working in Progress: 3D Hand Tracking for Extended Reality,” by MSc student Zhen Xu and PhD student Guorui Lu, has been accepted for a poster at the Neuro-inspired Computational Elements (NICE) 2025 conference in Heidelberg, Germany.
Our work “DeltaKWS: A 65nm 36nJ/Decision Bio-inspired Temporal-Sparsity-Aware Digital Keyword Spotting IC with 0.6V Near-Threshold SRAM,” together with ETH Zurich, University of Zurich, TU Delft, and Aalto University, was accepted by IEEE Transactions on Circuits and Systems for Artificial Intelligence.
Qinyu Chen presented at the TSMC OIP Ecosystem Forum 2024 in Amsterdam. Alongside our partnering professor Chang Gao at TU Delft and Professor Kwantae Kim at Aalto University, we shared our latest work in low-power IC design leveraging TSMC’s technologies.
At the National Open Science Festival in Maastricht, Qinyu Chen, alongside Guangzhi Tang (Maastricht University), Amirreza Yousefzadeh (University of Twente), and Manolis Sifalakis (imec), led a panel discussion on integrating open-science principles into AI and neuromorphic research during the Neuromorphic Computing session.
Presentation on the Event-Based Eye Tracking System at the Neuromorphic Computing Netherlands 2024 workshop in Eindhoven, the Netherlands.
Our work “FACET: Fast and Accurate Event-Based Eye Tracking Using Ellipse Modeling for Extended Reality” is online.
Our work “Four-class EEG Classification for Seizure Prediction and Detection Using a Lightweight CNN-LSTM” was accepted by BioCAS 2024.
We are happy to announce the success of our Event-based Eye Tracking Challenge at the CVPR AI4Streaming workshop. Our challenge report is now online. The corresponding 3ET+ dataset is now publicly available. Feel free to check out and follow our GitHub repository for more details and updates.
Our work “Epilepsy Seizure Detection and Prediction using an Approximate Spiking Convolutional Transformer” received the Best Paper Award — Honorable Mention from the IEEE CASS Neural Systems and Applications Technical Committee (NSATC) at ISCAS 2024 in Singapore.
Joined Leiden University as an Assistant Professor and very much enjoy the charm and rich history of the beautiful city of Leiden.
The Event-based Eye Tracking Challenge, a highlight of the AI for Streaming Workshop at CVPR 2024, has announced the successful completion of its first phase. The challenge is organized by the Institute of Neuroinformatics, University of Zurich and ETH Zurich, TU Delft, University of Würzburg, and Leiden University, and sponsored by Meta Reality Labs, Meta, Netflix, Sony Interactive Entertainment (FTG), and the University of Würzburg Computer Vision Lab.
Efficient Intelligence Group