Science relies on knowledge, knowledge relies on data. However, not all data is necessary, nor is it always managed efficiently or effectively. This is witnessed by the rapid growth of energy use in data centres with unsustainable socio-environmental effects. Industry has started creating and adopting energy-aware data management tactics. Science must do the same.To this aim, we build upon the methods and results emerged from the pragmatism of industrial practice and the rigour of academic research. We create an integrated `modeling-and-measurement’ method coming from both worlds, with energy-aware tactics learned in industry, and adapted and measured in academic research. [See the announcement here.]

Deep learning algorithms have shown strong performance in detecting and classifying arrhythmia, but their complexity and high power consumption limit their use in real-time wearable medical devices, where low-power and efficient solutions are essential. Approximate computing is a design approach that intentionally allows small, controlled errors in computation to reduce energy consumption, latency, and hardware complexity while maintaining output accuracy that is sufficient for the intended application. This project explores the use of approximate computing to reduce the energy usage of deep learning models for arrhythmia detection.



</a>

Innoguard targets developing novel methods for the quality assurance of autonomous cyber-physical systems by creating a tailored training program for early-stage researchers, with scientific objectives including methods to automate quality assessment and behavior evolution using AI techniques and enhancing dependability through real-time security, privacy, and uncertainty handling solutions. Additionally, InnoGuard seeks to improve the trustworthiness of AI methods, enhance environmental sustainability by increasing the energy efficiency of cyber-physical systems with the usage of AI methods, including large language models, and validate such techniques in open-source contexts. [Read more.]

In this project, we aim to answer the fundamental questions about how sustainability is affected by the way software running in the cloud is structured, and in doing so, develop tools that help making software more sustainable. [Read more.]

Green Lab