Skip to main content
Project

Geometric AI - Opening the black box of neural networks for responsible intelligence

Tampere University
Duration of project1.1.2026–30.6.2027
Area of focusTechnology

Although deep neural networks have shown success in various fields, they remain largely a mystery and are often referred to as black boxes, since our understanding of the fundamental mechanisms guiding their learning and predictions is still limited. The project prepares the commercialization of AI modules related to AI decision-making and operation, which help to demystify the black-box nature of current neural networks and also provide new perspectives on how to develop energy-efficient, privacy-preserving, explainable, and responsible AI models (aligned with human values and actions).