Reduce and explain BERT. New green AI technique improves efficiency of language models

A team of researchers at the University of Granada has developed an innovative methodology for the compression of BERT-based language models. The approach, called Persistent BERT Compression and Explainability (PBCE), uses persistent homology to identify and eliminate redundant neurons, achieving a reduction in model size of up to 47% in BERT Base and 42% in BERT Large without significantly affecting the accuracy of natural language processing (NLP) tasks.

“Health & AI” Workshop ENIA IAFER Project

We are pleased to announce the organisation of the free Workshop ‘Health and AI’, on the increasing importance that Artificial Intelligence, AI, plays in Health. This activity is part of the Artificial Intelligence, Ethical, Responsible and General Purpose (IAFER) project. It will be held in person at the UGR-AI Building in the PTS on Tuesday […]

11 FEB – Day of Women and Girls in Science. DaSCI Edition 2025

On 11 February we commemorate the International Day of Women and Girls in Science, a key day to highlight the role of women in science and technology, as well as to promote equal opportunities in these disciplines. In Spain, the central theme this year will be mental health, a fundamental aspect both for the well-being […]

Researchers from our centre awarded the ‘Juan Antonio García Torres’ prize

The prestigious ‘Juan Antonio García Torres’ prize, organised by the Royal Academy of Medicine and Surgery of Eastern Andalusia in collaboration with the Official College of Physicians of Granada, has recently been awarded to the work entitled ‘Identification of novel biomarkers in the early diagnosis of malignant melanoma by untargeted liquid chromatography coupled to high-resolution […]