DeepEn2023

DeepEn2023是由乔治亚州立大学和北卡罗来纳大学夏洛特分校的研究团队创建的大型能源数据集，专注于边缘人工智能系统的能源效率。该数据集覆盖了多种内核、深度神经网络模型及流行的边缘AI应用，旨在提高设备深度学习在边缘AI系统中的可持续性透明度。数据集内容包括数千个TensorFlow Lite模型，这些模型在不同的硬件平台上执行，以创建全面的能源消耗数据。创建过程涉及使用Monsoon Power Monitor捕捉模型执行期间的电力消耗数据，并将其转换为碳排放估计。DeepEn2023的应用领域主要集中在通过优化AI系统的能源效率来减少碳排放，支持全球气候变化的缓解。

DeepEn2023 is a large-scale energy dataset developed by research teams from Georgia State University and the University of North Carolina at Charlotte, focusing on the energy efficiency of edge artificial intelligence systems. This dataset covers various kernels, deep neural network models and popular edge AI applications, aiming to improve the sustainability and transparency of on-device deep learning in edge AI systems. The dataset includes thousands of TensorFlow Lite models, which are executed on different hardware platforms to generate comprehensive energy consumption data. The construction of DeepEn2023 involves using the Monsoon Power Monitor to capture power consumption data during model execution, and converting such data into carbon emission estimates. The main application fields of DeepEn2023 center on reducing carbon emissions by optimizing the energy efficiency of AI systems, to support global climate change mitigation.