Towards Planetary Scale Machine Learning for Everyone
Abstract
In the age of machine learning as a service, the earth science community has faced the challenge of adapting algorithms built for, among other things, phone camera imagery and language processing to answer scientific questions. We present a novel layer of abstraction built for earth and remote sensing scientists into the well-known Google Earth Engine platform with Google Cloud AI Platform and explore its application to global scale, deep-learning powered analysis. We will go on to discuss the use of modern machine-learning practices and unsupervised learning that, when paired with Earth Engine, decrease the burden of label collection while achieving parity with human perception on global land condition mapping tasks with uncertainties.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2020
- Bibcode:
- 2020AGUFMIN024..06B
- Keywords:
-
- 1908 Cyberinfrastructure;
- INFORMATICS;
- 1932 High-performance computing;
- INFORMATICS;
- 1942 Machine learning;
- INFORMATICS;
- 1976 Software tools and services;
- INFORMATICS