Aarhus Universitets segl


The Tiny Machine Learning (TinyML) Research Group is at the intersection of applied machine learning, embedded systems development, signal processing, and IoT. Modern microcontrollers are rather energy-efficient and allows for efficient data acquisition, signal processing and processing of machine learning models. Our project work includes design and application of classification, regression and anomaly detection in IoT-like devices. It also involves the design of Neural Network Accelerators that utilize both Artificial Neural Networks and Spiking Neural Networks. These efforts are aimed at augmenting the functionality and capability of low-power, intelligent devices.


TinyML enables significant processing and inference to be performed directly on devices, and by using power to compute rather than to transmit data, a significant increase in battery life, and corresponding reduction in required data can often be achieved. This approach is particularly beneficial for battery-powered devices.

In addition to research and development, the group offers an elective course in TinyML. This course aims to provide students with an understanding of TinyML's application across various disciplines, preparing them for future innovations in energy-efficient technology.