Summary of Cudle: Learning Under Label Scarcity to Detect Cannabis Use in Uncontrolled Environments, by Reza Rahimi Azghan et al.
CUDLE: Learning Under Label Scarcity to Detect Cannabis Use in Uncontrolled Environments
by Reza Rahimi Azghan, Nicholas C. Glodosky, Ramesh Kumar Sah, Carrie Cuttler, Ryan McLaughlin, Michael J. Cleveland, Hassan Ghasemzadeh
First submitted to arxiv on: 4 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 framework called CUDLE (Cannabis Use Detection with Label Efficiency) is proposed to automatically detect cannabis consumption in free-living environments. This framework leverages self-supervised learning with real-world wearable sensor data to overcome the challenges of obtaining accurate labels. The approach consists of a contrastive learning framework that identifies cannabis consumption moments using sensor-derived data. A self-supervised pretext task is first used to learn robust representations, which are then fine-tuned in a downstream task with a shallow classifier. This enables CUDLE to outperform traditional supervised methods, especially when limited labeled data is available. The approach is evaluated through a clinical study with 20 cannabis users, collecting over 500 hours of wearable sensor data alongside user-reported cannabis use moments. The results show that CUDLE achieves a higher accuracy of 73.4% compared to 71.1% for the supervised approach, while using 75% less labels. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Cannabis use detection is an important healthcare challenge. Researchers have developed a new way to do this using wearable sensors and machine learning. The method, called CUDLE, can detect when someone has used cannabis without needing them to tell us first. This is helpful because it’s hard to get accurate information about cannabis use in real-life situations. CUDLE uses special algorithms that learn from the sensor data and don’t need labels to do this. It was tested with 20 people who use cannabis, and it worked better than other methods when there wasn’t much labeled data. |
Keywords
* Artificial intelligence * Machine learning * Self supervised * Supervised
