The Glasgow Intelligent Computing Laboratory (gicLAB for short), is a research group in the School of Computing Science at the University of Glasgow. We are part of the Glasgow Systems Section (GLASS), where we focus on research at the intersection of Computing Systems and Machine Learning.
For the latest updates, our Twitter is @gic_lab, and our GitHub is @gicLAB. You can also learn more about our publications and people on this site.
Our paper “Improving Robustness Against Adversarial Attacks with Deeply Quantized Neural Networks” has been accepted at International Joint Conference on Neural Networks (IJCNN) 2023. This work is a collaboration with our colleagues at STMicroelectronics and the first publication of our PETRAS project MAISE (Multimodal AI based Security at the Edge).... [Read More]
gicLAB presents Transfer-Tuning in AccML@HiPEAC
Our student Perry Gibson gave a short invited talk on Transfer-Tuning at the Accelerated Machine Learning (AccML) workshop, in HiPEAC 2023, Tolousse, France.
Paper published in NeurIPS ML Safety Workshop
Nick Louloudakis (based at the University of Edinburgh) published a paper at the NeurIPS ML Safety Workshop 2022, called “Assessing Robustness of Image Recognition Models to Changes in the Computational Environment”. [Read More]
SECDA-TFLite Published in JPDC
Our paper “SECDA-TFLite: A Toolkit for Efficient Development of FPGA-based DNN Accelerators for Edge Inference” has been accepted at Elsevier Journal of Parallel and Distributed Computing (JPDC). In addition, our code is available as an open-source toolkit for the community.
Transfer-Tuning Published at PACT 2022
Our paper Transfer-Tuning was published at PACT 2022, along with an artifact review. The preprint is available on arXiv. See paper details in our publication list.
Bifrost Published at IEEE ISPASS 2022
Congratulations to first author Axel Stjerngren
See paper details in our publication list.
gicLAB receives funding for BonsAPPs AIMDDE project
gicLAB is undertaking a 2 month project for an AI in industry challenge [Read More]
SECDA Published at IEEE SBAC-PAD 2021
Congratulations to first author Jude Haris
Publication: SECDA: Efficient Hardware/Software Co-Design of FPGA-based DNN Accelerators for Edge Inference @ SBAC-PAD 2021. [IEEE paper] [arXiv] [Blogpost]