Summary of Pepr: Performance Per Resource Unit As a Metric to Promote Small-scale Deep Learning in Medical Image Analysis, by Raghavendra Selvan et al.
PePR: Performance Per Resource Unit as a Metric to Promote Small-Scale Deep Learning in Medical Image Analysis
by Raghavendra Selvan, Bob Pepin, Christian Igel, Gabrielle Samuel, Erik B Dam
First submitted to arxiv on: 19 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary A novel approach is proposed to address the growing concern of large-scale deep learning (DL) models’ resource-intensive nature, which hinders researchers’ access to resources in the Global South. The study analyzes a diverse family of 131 DL architectures for medical image analysis tasks, showcasing their performance per resource unit (PePR score). By evaluating trends on three datasets, the authors argue that small-scale, specialized models are more effective than striving for large-scale ones. Additionally, fine-tuning existing pretrained models can significantly reduce computational resources and data required compared to training from scratch. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps make deep learning more accessible to everyone by looking at how big AI models use up resources like computers, energy, and carbon emissions. It shows that using smaller, specialized models is often better than trying to make bigger ones. The study also finds that using pre-trained models and fine-tuning them for new tasks can save a lot of computer time and data space. |
Keywords
* Artificial intelligence * Deep learning * Fine tuning
