TinyML-UK Network

The development trajectory of large AI models often follows a "bigger is better" approach, focusing on increased parameters, storage, compute power, and bandwidth to enhance accuracy and results. However, a counter-movement known as "Tiny Machine Learning" (tinyML) is gaining traction. tinyML optimizes technology for deployment on low-cost, edge computing devices, enabling efficient machine learning algorithms that can operate over long periods on battery power. By processing data locally, tinyML offers advantages in low latency, privacy preserving, and the potential for real-time interventions and faster response times. By democratizing AI, tinyML makes advanced capabilities accessible to a broader range of developers and end-users, unlocking applications in areas such as wildlife monitoring and medical wearable devices for personalized health management. Despite the UK's significant investment in AI, efforts specifically focused on AI hardware, including tinyML, remain limited. The UK's ability to compete globally in this field is hindered by a fragmented research community. To address this, the tinyML-UK Network will foster interdisciplinary collaboration between hardware designers and AI developers, bridging the gap between these traditionally separate domains. The network's vision is to bring together experts from various domains, including AI model hardware accelerators, neuromorphic computing, software, system tools for edge devices, chip design, and custom-built AI hardware. By establishing a common language and facilitating interaction between engineers and AI scientists, the network will drive innovation and shape the roadmap for AI and edge computing in the UK. And worldwide. To achieve its goals, the network will undertake a comprehensive program of activities. These include networking, training, competitions, and technical knowledge exchange events, building on the successes of the tinyML Foundation. Key events will include an initial kick-off event, an annual conference, four scientific workshops, three stakeholder events, specialized tutorial sessions, and capacity-building activities. These events will foster industry-academic partnerships, support early-career researchers, and promote the exchange of knowledge and ideas across disciplines. The network will also leverage international industry collaborations to advance tinyML research and encourage cross-sector exchanges. This will include organizing industry visits, initiating an Industry Mentorship Program, and promoting academic-industry publications. Engagement with stakeholders will be crucial for exploring real-world applications of tinyML in areas such as sustainability, health and well-being, and security. The network will run sector-specific workshops, policy forums, and impact briefings to inform and involve decision-makers. Outreach and dissemination efforts will include a dedicated website, newsletters, and social media engagement. The network will also promote its activities through the tinyML Foundation's channels, ensuring broad visibility and impact. By the end of the network period, we expect to have a self-sustaining momentum of internationally coordinated research, incorporating the initial investigators, but extending beyond to new groups and the network of SMEs developing in this area. The tinyML-UK Network has the potential to place the UK's expertise at the centre of this emerging technology, revolutionizing the way we process data and deliver low-energy edge computation. Through interdisciplinary collaboration, the network will address key challenges and opportunities, ensuring the UK's competitiveness in the rapidly evolving field of AI.