UK Supercomputing Initiative to Enhance Emotion Recognition with Synthetic Data
Goldsmiths University is utilizing UK supercomputing resources to develop a facial emotion recognition system using synthetic data. The project, part of the AI Research Resource (AIRR), aims to combat biases in existing emotion recognition technologies.
Goldsmiths University has initiated a three-month study leveraging supercomputing power from the University of Cambridge's Dawn and the University of Bristol's Isambard-AI to improve facial emotion recognition systems. The project, supported by the UK's AI Research Resource (AIRR), focuses on training algorithms using synthetically generated images to create a diverse and well-labeled dataset.
This initiative addresses the challenges of bias in existing systems, which often misclassify emotions from various demographic groups due to poorly labeled datasets. AIRR, launched in July 2022, aims to significantly enhance the UK's AI computing capacity by investing £1 billion by 2030.
The Isambard-AI is currently the most powerful public computing facility in the UK, while the Dawn supercomputer features over 1,000 Intel GPUs. Other projects utilizing AIRR resources include AI for dementia assistance and cancer detection through MRI analysis.
Comments