What is it about?
This paper recommends the creation of detailed documentation for energy datasets, including information about the demographics of people likely to be affected by the machine learning outcomes that are based on that data. We show that a documentation tool developed in the computer vision and natural language processing communities applies neatly to datasets in the energy domain.
Featured Image
Photo by Sigmund on Unsplash
Why is it important?
The energy sector is increasingly reliant on machine learning to inform its decision making. In order for those decisions to be made in an ethical manner, it is crucial that developers and users of those machine learning algorithms understand the data that formed the models and outcomes.
Perspectives
In many areas of machine learning, it is important to obscure the demographics of the people involved, so that outcomes are fair, and based on the features we believe to be relevant, rather than on demographic classes. But in decisions about energy innovations, equitable outcomes rely on understanding demographics, not masking them.
Ilana Heintz
Synoptic Engineering
Read the Original
This page is a summary of: Datasheets for Energy Datasets: An Ethically-Minded Approach to Documentation, June 2023, ACM (Association for Computing Machinery),
DOI: 10.1145/3599733.3600249.
You can read the full text:
Contributors
The following have contributed to this page
