The increasing energy consumption and carbon footprint of deep learning (DL) due to growing compute requirements has become a cause of concern. In this work, we focus on the carbon footprint of developing DL models for medical image analysis (MIA), where volumetric images of high spatial resolution are handled. In this study, we present and compare the features of four tools from literature to quantify the carbon footprint of DL. Using one of these tools we estimate the carbon footprint of medical image segmentation pipelines. We choose nnU-net as the proxy for a medical image segmentation pipeline and experiment on three common datasets. With our work we hope to inform on the increasing energy costs incurred by MIA. We discuss simple strategies to cut-down the environmental impact that can make model selection and training processes more efficient.

Artificial intelligence is an increasingly important element of science, medicine, and even the minutiae of our daily lives. Chatbots, digital assistants, and movie and music recommendations from streaming services all depend on "deep learning"--a process by which computer models are trained to recognize patterns in data. That training requires powerful computers and lots and lots of energy--and associated carbon emissions. One of the most elaborate deep learning models, designed to produce human-like language and known as GPT-3, requires an amount of energy equivalent to the yearly consumption of 126 Danish homes and creates a carbon footprint equivalent to traveling 700,000 kilometers by car for a single training session. Still, the computing power used in deep learning grew 300,000-fold between 2012 and 2018, and if that pace of growth continues it's not hard to see how artificial intelligence could have a major climate impact.

Deep-learning A.I. is the machine learning technology that powers everything from cutting-edge natural language processing to machine vision tools. It may also be powering climate change -- as a result of the massive energy consumption and CO2 emissions associated with training these deep-learning models. As the use of deep learning has exploded, so has the compute power associated with them, although this effect is rarely studied. Researchers at the University of Copenhagen's Department of Computer Science are working to change that, however. They've developed a tool called Carbontracker, which works out the energy consumption associated with deep-learning algorithms and then converts this into a prediction about CO2 emissions.

However, the rapidly evolving technology, one that has otherwise been expected to serve as an effective weapon against climate change, has a downside that many people are unaware of -- sky high energy consumption. Artificial intelligence, and particularly the subfield of deep learning, appears likely to become a significant climate culprit should industry trends continue. In only six years -- from 2012 to 2018 -- the compute needed for deep learning has grown 300,000%. However, the energy consumption and carbon footprint associated with developing algorithms is rarely measured, despite numerous studies that clearly demonstrate the growing problem. In response to the problem, two students at the University of Copenhagen's Department of Computer Science, Lasse F. Wolff Anthony and Benjamin Kanding, together with Assistant Professor Raghavendra Selvan, have developed a software programme they call Carbontracker.

The power and energy monitoring in carbontracker is limited to a few main components of computational systems. Additional power consumed by the supporting infrastructure, such as that used for cooling or power delivery, is accounted for by multiplying the measured power by the pue of the data center hosting the compute, as suggested by Strubell2019. Previous research has examined pue and its shortcomings (Yuventi2013). These shortcomings may largely be resolved by data centers reporting an average pue instead of a minimum observed value. In our work, we use a pue of 1.58, the global average for data centers in 2018 as reported by Ascierto2018.222Early

Deep learning (DL) can achieve impressive results across a wide variety of tasks, but this often comes at the cost of training models for extensive periods on specialized hardware accelerators. This energy-intensive workload has seen immense growth in recent years. Machine learning (ML) may become a significant contributor to climate change if this exponential trend continues. If practitioners are aware of their energy and carbon footprint, then they may actively take steps to reduce it whenever possible. In this work, we present Carbontracker, a tool for tracking and predicting the energy and carbon footprint of training DL models. We propose that energy and carbon footprint of model development and training is reported alongside performance metrics using tools like Carbontracker. We hope this will promote responsible computing in ML and encourage research into energy-efficient deep neural networks.