Patricia Lago

Ivano Malavolta, Vincenzo Stoico, Patricia Lago

In this chapter we share our experience in running ten editions of the Green Lab course at the Vrije Universiteit Amsterdam, the Netherlands. The course is given in the Software Engineering and Green IT track of the Computer Science Master program of the VU. The course takes place every year over a 2-month period and teaches Computer Science students the fundamentals of Empirical Software Engineering in the context of energy-efficient software. The peculiarity of the course is its research orientation: at the beginning of the course the instructor presents a catalog of scientifically relevant goals, and each team of students signs up for one of them and works together for 2 months on their own experiment for achieving the goal. Each team goes over the classic steps of an empirical study, starting from a precise formulation of the goal and research questions to context definition, selection of experimental subjects and objects, definition of experimental variables, experiment execution, data analysis, and reporting. Over the years, the course became well-known within the Software Engineering community since it led to several scientific studies that have been published at various scientific conferences and journals. Also, students execute their experiments using \textit{open-source tools}, which are developed and maintained by researchers and other students within the program, thus creating a virtuous community of learners where students exchange ideas, help each other, and learn how to collaboratively contribute to open-source projects in a safe environment.

Rafiullah Omar, Justus Bogner, Henry Muccini et al.

Background: Machine learning (ML) model composition is a popular technique to mitigate shortcomings of a single ML model and to design more effective ML-enabled systems. While ensemble learning, i.e., forwarding the same request to several models and fusing their predictions, has been studied extensively for accuracy, we have insufficient knowledge about how to design energy-efficient ensembles. Objective: We therefore analyzed three types of design decisions for ensemble learning regarding a potential trade-off between accuracy and energy consumption: a) ensemble size, i.e., the number of models in the ensemble, b) fusion methods (majority voting vs. a meta-model), and c) partitioning methods (whole-dataset vs. subset-based training). Methods: By combining four popular ML algorithms for classification in different ensembles, we conducted a full factorial experiment with 11 ensembles x 4 datasets x 2 fusion methods x 2 partitioning methods (176 combinations). For each combination, we measured accuracy (F1-score) and energy consumption in J (for both training and inference). Results: While a larger ensemble size significantly increased energy consumption (size 2 ensembles consumed 37.49% less energy than size 3 ensembles, which in turn consumed 26.96% less energy than the size 4 ensembles), it did not significantly increase accuracy. Furthermore, majority voting outperformed meta-model fusion both in terms of accuracy (Cohen's d of 0.38) and energy consumption (Cohen's d of 0.92). Lastly, subset-based training led to significantly lower energy consumption (Cohen's d of 0.91), while training on the whole dataset did not increase accuracy significantly. Conclusions: From a Green AI perspective, we recommend designing ensembles of small size (2 or maximum 3 models), using subset-based training, majority voting, and energy-efficient ML algorithms like decision trees, Naive Bayes, or KNN.

Iffat Fatima, Markus Funke, Patricia Lago

Fairness issues often remain hidden in digital systems, making them difficult to detect and even more difficult to address. In this study, we introduce a fairness-first Design Thinking (DT) approach to support addressing fairness concerns in software architecture (SA) design. We implemented our approach in a graduate-level course where students executed all steps of our DT approach as part of an assignment. We analyzed the assignment data to reflect on the implications for applying the DT approach in SA and teaching the DT approach in SA education. As a result of this study, we provide (i) a DT approach for SA, (ii) implications of the DT approach on handling fairness in both problem and solution spaces, and (iii) implications for education. Our reflections highlight that fairness theory and context identification are essential for a holistic, fairness-first design. We propose the use of composite views to address cross-cutting concerns such as fairness. In the future, we will update the course material to provide end-to-end fairness traceability in SA, helping students to understand how fairness concerns can be translated into actionable design decisions.

Heli Järvenpää, Patricia Lago, Justus Bogner et al.

The rapid adoption of artificial intelligence (AI) and machine learning (ML) has generated growing interest in understanding their environmental impact and the challenges associated with designing environmentally friendly ML-enabled systems. While Green AI research, i.e., research that tries to minimize the energy footprint of AI, is receiving increasing attention, very few concrete guidelines are available on how ML-enabled systems can be designed to be more environmentally sustainable. In this paper, we provide a catalog of 30 green architectural tactics for ML-enabled systems to fill this gap. An architectural tactic is a high-level design technique to improve software quality, in our case environmental sustainability. We derived the tactics from the analysis of 51 peer-reviewed publications that primarily explore Green AI, and validated them using a focus group approach with three experts. The 30 tactics we identified are aimed to serve as an initial reference guide for further exploration into Green AI from a software engineering perspective, and assist in designing sustainable ML-enabled systems. To enhance transparency and facilitate their widespread use and extension, we make the tactics available online in easily consumable formats. Wide-spread adoption of these tactics has the potential to substantially reduce the societal impact of ML-enabled systems regarding their energy and carbon footprint.

Francisco Durán, Silverio Martínez-Fernández, Matias Martinez et al.

The growing use of large machine learning models highlights concerns about their increasing computational demands. While the energy consumption of their training phase has received attention, fewer works have considered the inference phase. For ML inference, the binding of ML models to the ML system for user access, known as ML serving, is a critical yet understudied step for achieving efficiency in ML applications. We examine the literature in ML architectural design decisions and Green AI, with a special focus on ML serving. The aim is to analyze ML serving architectural design decisions for the purpose of understanding and identifying them with respect to quality characteristics from the point of view of researchers and practitioners in the context of ML serving literature. Our results (i) identify ML serving architectural design decisions along with their corresponding components and associated technological stack, and (ii) provide an overview of the quality characteristics studied in the literature, including energy efficiency. This preliminary study is the first step in our goal to achieve green ML serving. Our analysis may aid ML researchers and practitioners in making green-aware architecture design decisions when serving their models.

Luís Cruz, João Paulo Fernandes, Maja H. Kirkeby et al.

The environmental impact of Artificial Intelligence (AI)-enabled systems is increasing rapidly, and software engineering plays a critical role in developing sustainable solutions. The "Greening AI with Software Engineering" CECAM-Lorentz workshop (no. 1358, 2025) funded by the Centre Européen de Calcul Atomique et Moléculaire and the Lorentz Center, provided an interdisciplinary forum for 29 participants, from practitioners to academics, to share knowledge, ideas, practices, and current results dedicated to advancing green software and AI research. The workshop was held February 3-7, 2025, in Lausanne, Switzerland. Through keynotes, flash talks, and collaborative discussions, participants identified and prioritized key challenges for the field. These included energy assessment and standardization, benchmarking practices, sustainability-aware architectures, runtime adaptation, empirical methodologies, and education. This report presents a research agenda emerging from the workshop, outlining open research directions and practical recommendations to guide the development of environmentally sustainable AI-enabled systems rooted in software engineering principles.

Rafiullah Omar, Justus Bogner, Joran Leest et al.

ML-enabled systems that are deployed in a production environment typically suffer from decaying model prediction quality through concept drift, i.e., a gradual change in the statistical characteristics of a certain real-world domain. To combat this, a simple solution is to periodically retrain ML models, which unfortunately can consume a lot of energy. One recommended tactic to improve energy efficiency is therefore to systematically monitor the level of concept drift and only retrain when it becomes unavoidable. Different methods are available to do this, but we know very little about their concrete impact on the tradeoff between accuracy and energy efficiency, as these methods also consume energy themselves. To address this, we therefore conducted a controlled experiment to study the accuracy vs. energy efficiency tradeoff of seven common methods for concept drift detection. We used five synthetic datasets, each in a version with abrupt and one with gradual drift, and trained six different ML models as base classifiers. Based on a full factorial design, we tested 420 combinations (7 drift detectors * 5 datasets * 2 types of drift * 6 base classifiers) and compared energy consumption and drift detection accuracy. Our results indicate that there are three types of detectors: a) detectors that sacrifice energy efficiency for detection accuracy (KSWIN), b) balanced detectors that consume low to medium energy with good accuracy (HDDM_W, ADWIN), and c) detectors that consume very little energy but are unusable in practice due to very poor accuracy (HDDM_A, PageHinkley, DDM, EDDM). By providing rich evidence for this energy efficiency tactic, our findings support ML practitioners in choosing the best suited method of concept drift detection for their ML-enabled systems.

Apoorva Nalini Pradeep Kumar, Justus Bogner, Markus Funke et al.

Recent advances in artificial intelligence (AI) capabilities have increased the eagerness of companies to integrate AI into software systems. While AI can be used to have a positive impact on several dimensions of sustainability, this is often overshadowed by its potential negative influence. While many studies have explored sustainability factors in isolation, there is insufficient holistic coverage of potential sustainability benefits or costs that practitioners need to consider during decision-making for AI adoption. We therefore aim to synthesize trade-offs related to sustainability in the context of integrating AI into software systems. We want to make the sustainability benefits and costs of integrating AI more transparent and accessible for practitioners. The study was conducted in collaboration with a Dutch financial organization. We first performed a rapid review that led to the inclusion of 151 research papers. Afterward, we conducted six semi-structured interviews to enrich the data with industry perspectives. The combined results showcase the potential sustainability benefits and costs of integrating AI. The labels synthesized from the review regarding potential sustainability benefits were clustered into 16 themes, with "energy management" being the most frequently mentioned one. 11 themes were identified in the interviews, with the top mentioned theme being "employee wellbeing". Regarding sustainability costs, the review discovered seven themes, with "deployment issues" being the most popular one, followed by "ethics & society". "Environmental issues" was the top theme from the interviews. Our results provide valuable insights to organizations and practitioners for understanding the potential sustainability implications of adopting AI.

Klervie Toczé, Andrei Dragomir, Vincenzo Stoico et al.

Enabling observability in software systems brings many benefits. It can, for example, ease the identification of issues or the implementation of improvements. It is especially critical to be able to observe sustainability-related dimensions of systems to know and mitigate their impact. However, adding observability to a system, especially related to software sustainability, requires technical knowledge that may not be available in every project that would benefit from it. In this work, we propose an architectural blueprint along with its deployment code that can be used to facilitate the addition of observability in software systems. As a special case, it includes measuring the energy consumption of software. This toolkit provides support in defining which components are necessary for a given use case and for structuring their deployment. Moreover, we exemplify the addition of observability in two different use cases.

Francisco Durán, Matias Martinez, Patricia Lago et al.

The rapid growth of language models, particularly in code generation, requires substantial computational resources, raising concerns about energy consumption and environmental impact. Optimizing language models inference resource utilization is crucial, and Small Language Models (SLMs) offer a promising solution to reduce resource demands. Our goal is to analyze the impact of deep learning serving configurations, defined as combinations of runtime engines and execution providers, on resource utilization, in terms of energy consumption, execution time, and computing-resource utilization from the point of view of software engineers conducting inference in the context of code generation SLMs. We conducted a technology-oriented, multi-stage experimental pipeline using twelve code generation SLMs to investigate energy consumption, execution time, and computing-resource utilization across the configurations. Significant differences emerged across configurations. CUDA execution provider configurations outperformed CPU execution provider configurations in both energy consumption and execution time. Among the configurations, TORCH paired with CUDA demonstrated the greatest energy efficiency, achieving energy savings from 37.99% up to 89.16% compared to other serving configurations. Similarly, optimized runtime engines like ONNX with the CPU execution provider achieved from 8.98% up to 72.04% energy savings within CPU-based configurations. Also, TORCH paired with CUDA exhibited efficient computing-resource utilization. Serving configuration choice significantly impacts resource utilization. While further research is needed, we recommend the above configurations best suited to software engineers' requirements for enhancing serving resource utilization efficiency.

Joran Leest, Claudia Raibulet, Ilias Gerostathopoulos et al.

We propose Expert Monitoring, an approach that leverages domain expertise to enhance the detection and mitigation of concept drift in machine learning (ML) models. Our approach supports practitioners by consolidating domain expertise related to concept drift-inducing events, making this expertise accessible to on-call personnel, and enabling automatic adaptability with expert oversight.

Michel Albonico, Ivano Malavolta, Gustavo Pinto et al.

Robots are becoming more and more commonplace in many industry settings. This successful adoption can be partly attributed to (1) their increasingly affordable cost and (2) the possibility of developing intelligent, software-driven robots. Unfortunately, robotics software consumes significant amounts of energy. Moreover, robots are often battery-driven, meaning that even a small energy improvement can help reduce its energy footprint and increase its autonomy and user experience. In this paper, we study the Robot Operating System (ROS) ecosystem, the de-facto standard for developing and prototyping robotics software. We analyze 527 energy-related data points (including commits, pull-requests, and issues on ROS-related repositories, ROS-related questions on StackOverflow, ROS Discourse, ROS Answers, and the official ROS Wiki). Our results include a quantification of the interest of roboticists on software energy efficiency, 10 recurrent causes, and 14 solutions of energy-related issues, and their implied trade-offs with respect to other quality attributes. Those contributions support roboticists and researchers towards having energy-efficient software in future robotics projects.

Ilias Gerostathopoulos, Thomas Vogel, Danny Weyns et al.

With the increase of research in self-adaptive systems, there is a need to better understand the way research contributions are evaluated. Such insights will support researchers to better compare new findings when developing new knowledge for the community. However, so far there is no clear overview of how evaluations are performed in self-adaptive systems. To address this gap, we conduct a mapping study. The study focuses on experimental evaluations published in the last decade at the prime venue of research in software engineering for self-adaptive systems -- the International Symposium on Software Engineering for Adaptive and Self-Managing Systems (SEAMS). Results point out that specifics of self-adaptive systems require special attention in the experimental process, including the distinction of the managing system (i.e., the target of evaluation) and the managed system, the presence of uncertainties that affect the system behavior and hence need to be taken into account in data analysis, and the potential of managed systems to be reused across experiments, beyond replications. To conclude, we offer a set of suggestions derived from our study that can be used as input to enhance future experiments in self-adaptive systems.

Francesco Osborne, Henry Muccini, Patricia Lago et al.

Context. Systematic Reviews (SRs) are means for collecting and synthesizing evidence from the identification and analysis of relevant studies from multiple sources. To this aim, they use a well-defined methodology meant to mitigate the risks of biases and ensure repeatability for later updates. SRs, however, involve significant effort. Goal. The goal of this paper is to introduce a novel methodology that reduces the amount of manual tedious tasks involved in SRs while taking advantage of the value provided by human expertise. Method. Starting from current methodologies for SRs, we replaced the steps of keywording and data extraction with an automatic methodology for generating a domain ontology and classifying the primary studies. This methodology has been applied in the Software Engineering sub-area of Software Architecture and evaluated by human annotators. Results. The result is a novel Expert-Driven Automatic Methodology, EDAM, for assisting researchers in performing SRs. EDAM combines ontology-learning techniques and semantic technologies with the human-in-the-loop. The first (thanks to automation) fosters scalability, objectivity, reproducibility and granularity of the studies; the second allows tailoring to the specific focus of the study at hand and knowledge reuse from domain experts. We evaluated EDAM on the field of Software Architecture against six senior researchers. As a result, we found that the performance of the senior researchers in classifying papers was not statistically significantly different from EDAM. Conclusions. Thanks to automation of the less-creative steps in SRs, our methodology allows researchers to skip the tedious tasks of keywording and manually classifying primary studies, thus freeing effort for the analysis and the discussion.

Henry Muccini, Patricia Lago, Karthik Vaidyanathan et al.

Software architecture (SA) is celebrating 25 years. This is so if we consider the seminal papers establishing SA as a distinct discipline and scientific publications that have identified cornerstones of both research and practice, like architecture views, architecture description languages, and architecture evaluation. With the pervasive use of cloud provisioning, the dynamic integration of multi-party distributed services, and the steep increase in the digitalization of business and society, making sound design decisions encompasses an increasingly-large and complex problem space. The role of SA is essential as never before, so much so that no organization undertakes `serious' projects without the support of suitable architecture practices. But, how did SA practice evolve in the past 25 years? and What are the challenges ahead? There have been various attempts to summarize the state of research and practice of SA. Still, we miss the practitioners' view on the questions above. To fill this gap, we have first extracted the top-10 topics resulting from the analysis of 5,622 scientific papers. Then, we have used such topics to design an online survey filled out by 57 SA practitioners with 5 to 20+ years of experience. We present the results of the survey with a special focus on the SA topics that SA practitioners perceive, in the past, present and future, as the most impactful. We finally use the results to draw preliminary takeaways.

Daniel S. Katz, Kyle E. Niemeyer, Sandra Gesing et al.

This report records and discusses the Fourth Workshop on Sustainable Software for Science: Practice and Experiences (WSSSPE4). The report includes a description of the keynote presentation of the workshop, the mission and vision statements that were drafted at the workshop and finalized shortly after it, a set of idea papers, position papers, experience papers, demos, and lightning talks, and a panel discussion. The main part of the report covers the set of working groups that formed during the meeting, and for each, discusses the participants, the objective and goal, and how the objective can be reached, along with contact information for readers who may want to join the group. Finally, we present results from a survey of the workshop attendees.