Call for Papers

Submission Deadline: May 15, 2026

Recent advances in AI, especially large language and multimodal models, have been driven by rapid scaling of data, models, and compute—but at growing energy, resource, and environmental costs that few labs can afford. This workshop focuses on the sustainability of AI systems across their lifecycle, emphasizing data-, compute-, and energy-efficient methods, as well as quantitative, reproducible metrics and benchmarks. We invite contributions on methods, metrics, and real-world applications (including edge and embedded deployments) that enable AI systems to be both powerful and sustainable. Papers in this workshop must describe high-quality, original research.

Topics of Interest

Topics of interest cover all aspects of efficient and sustainable AI including, but not limited to:

• Data- and label-efficient learning under resource constraints, including theoretical analyses and empirical studies
• Leveraging underutilized or unconventional data sources to reduce data and compute demands
• Contributions in high performance computing machine learning pipelines including energy-, resource-, hardware-, and time-efficient perception and inference
• Efficient multimodal AI, including but not limited to exploiting complementary modalities for efficiency (e.g., adaptive modality selection, sensor scheduling, shared representations)
• Sustainable re-use, adaptation, and life-cycle management of ML models, including fine-tuning of large foundation models, in-context and test-time learning, continual learning, and federated learning.
• Metrics, methodologies, and tools for measuring and evaluating efficiency and sustainability in ML (e.g., budgeted performance under energy, compute, or data constraints; carbon accounting; standardized reporting; benchmarks)
• Contributions in applications including design, deployment, and evaluation of efficient models on edge and mobile devices, for example case studies in industrial, agricultural, mobility, healthcare, and other real-world environments
• Sustainable hardware and materials, including algorithm–hardware co-design, low-power accelerators, and environmentally responsible hardware life-cycles
• Ethical, socio-economic, and socio-cultural aspects of sustainable AI, such as equity of access, environmental justice, governance, and policy implications

Submission Tracks

Full Papers (Archival, PMLR Proceedings)

Original research papers of up to 7 pages for the main content, plus up to 2 additional pages for references (9 pages total). Accepted full papers will be published in a dedicated volume of the Proceedings of Machine Learning Research (PMLR).

Acceptance Criteria: Novelty and relevance to efficient and sustainable AI, technical soundness and appropriate methodology (including evaluation), potential impact, as well as clarity and quality of presentation.

Short Papers (Non-Archival)

Short papers of up to 4 pages for the main content, plus 1 page for references (5 pages total). Short papers are intended for work-in-progress, promising preliminary results, and ideas that would benefit from feedback and discussion. Short papers will not appear in the archival proceedings, allowing authors to later submit extended versions to other conferences or journals.

Acceptance Criteria: Clarity and plausibility of the core idea or hypothesis, methodological soundness of first results or proof of concept, potential for impact and stimulating discussion at the workshop.

Position Papers (Non-Archival)

Position papers of up to 2 pages plus 1 page for references (3 pages total). Contributions may articulate research visions, identify open challenges, propose benchmarks or standards, or offer critical perspectives on efficient and sustainable AI. Position papers are non-archival and are primarily meant to serve as a basis for discussions and interactive sessions at the workshop.

Acceptance Criteria: Originality and thought-provoking perspectives, a well-argued and clearly articulated position grounded in relevant literature or practice, potential to catalyze discussion.

All submissions must follow the official SuRE workshop formatting guidelines and will undergo double-blind peer review. Download the author kit for templates and style files:
On GitHub  ·  Download as ZIP

Important Dates

Contact

In case of questions, contact us at: ✉️ sure-organizers@dfki.de