Summary of Ai in Space For Scientific Missions: Strategies For Minimizing Neural-network Model Upload, by Jonah Ekelund et al.
AI in Space for Scientific Missions: Strategies for Minimizing Neural-Network Model Upload
by Jonah Ekelund, Ricardo Vinuesa, Yuri Khotyaintsev, Pierre Henri, Gian Luca Delzanno, Stefano Markidis
First submitted to arxiv on: 20 Jun 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Instrumentation and Methods for Astrophysics (astro-ph.IM)
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 The proposed AI solution for space exploration could revolutionize decision-making processes onboard spacecraft, reducing reliance on ground control and predefined procedures. An onboard AI/ML Processing Unit would run an inference engine, using pre-installed parameters that can be updated by uploading trained models. However, satellite uplinks have limited bandwidth and transmissions can be costly. To address this, researchers evaluate the use of reduced-precision and bare-minimum neural networks to reduce upload time, demonstrating a 98.9% reduction in model size while maintaining accuracy above 94%. This could enable more efficient data collection and transmission during space missions. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary AI is helping space exploration by letting spacecraft make decisions without needing human control. An AI/ML Processing Unit would be on the spacecraft, running an engine that makes predictions based on pre-installed parameters. These can be updated from Earth, but it’s expensive and slow. To fix this, scientists are exploring smaller neural networks that still get good results. They tested reducing the size of a model by up to 98.9% while keeping its accuracy above 94%. This could make space missions more efficient. |
Keywords
» Artificial intelligence » Inference » Precision
