Green Software

Energy efficiency, Green software, Sustainable AI, Green microservices, Ensemble learning, Concept drift, Reinforcement learning, Feature selection

• Explore strategies for balancing performance and energy consumption in ML models and microservices.
• Managing concept drift in ML applications.
• Develop guidelines for designing energy-efficient microservices.
• Ensemble learning for energy efficiency
• Assess the role of reinforcement learning in optimizing configurations for sustainability.

Green Software encompasses initiatives aimed at enhancing the energy efficiency and sustainability of software systems. This area of research focuses on optimizing both machine learning (ML) models and microservices architectures to mitigate their environmental impact. By investigating strategies for energy-efficient design, we aim to balance performance with ecological considerations.

Our work examines various methodologies, including ensemble learning configurations and feature selection techniques, to optimize ML models. Additionally, we explore the implications of concept drift and its management to ensure models remain effective over time while conserving energy. In the context of microservices, we focus on developing guidelines for energy-efficient design and resource utilization, particularly through the use of green autoscalers.

Leveraging advanced techniques like reinforcement learning, we aim to dynamically adapt configurations across software systems, fostering sustainable practices in software development. Ultimately, our research seeks to contribute to the creation of environmentally responsible software solutions that minimize energy consumption while delivering high performance.