IBM's Foundation Models will analyze vast databases of NASA satellite images and geographic information, and will enable efficient searching of more than 300 academic articles on Earth, to help formulate insights and uncover new and valuable knowledge for researchers dealing with The most pressing climate challenges
The computing giant IBM and NASA's Marshall Space Flight Center
Announcing today a collaboration whose main purpose is to use IBM's artificial intelligence technologies to reach
For new research insights from NASA's vast scientific and geographic database. first
IBM Foundation Models will be used to process and land satellite images of a sphere
The land collected over the years.
Foundation Models are artificial intelligence models trained on very broad datasets that are not
They will be tagged, so that they can be focused on different types of insight formation tasks. in the last five years
Such models have rapidly advanced the field of natural language processing (NLP), and IBM is a global pioneer.
In the application of these models beyond the different use of language.
NASA has been making satellite observations of the Earth for many years, to allow scientists to monitor and study
the earth and various physical and geographical phenomena. But given the enormous size of the database,
New and innovative approaches are now required to extract insights and useful information from the data. Therefore, purpose
The collaboration between IBM and NASA is to provide a more efficient way for scientists and researchers to analyze and generate insights
of these large data stores, and IBM's Foundation Models have the potential to accelerate the
Analyzing the data and discovering new insights about the planet, and advancing more quickly the scientific understanding of
Climate-related phenomena.
As part of the collaboration, IBM and NASA plan to develop several new technologies to extract insights
From the observations of the earth. One project will train IBM's Foundation Model in the field of geography
The Harmonized Landsat Sentinel-2 dataset that records land changes observed by satellites
surrounding the earth, changes in the geographic footprint of phenomena such as natural disasters, yields
Crop rotation and wildlife habitats. This is a data set that includes petabytes of
satellite imagery and associated data, and the Foundation Model technology will help researchers analyze it and provide
Critical insights about the Earth's environmental systems.
The collaboration is also expected to yield a search capability in the entire database of scientific literature about the Earth that has been collected
by NASA. IBM has developed a natural language processing (NLP) model that has been trained on nearly 300 articles
academics dealing with earth sciences, and will facilitate the extraction of new knowledge in the field. This is one of the databases
the largest in the world, and the training of the system was carried out on Red's OpenShift software platform
Hat, with the help of PrimeQA - a multilingual and open source question and answer based solution. beyond its use
As a tool for researchers, the new language model that will "specialize" in earth sciences, can also be used for data management
scientific and process management at NASA.
"The Foundation Model technology has proven itself to be a success in natural language processing, and it's time to expand
it to new areas important to business and society," said Raghu Ganti, chief researcher at IBM. "You applied
Foundation Model on geographic data, processes that spread over time and scientific data types
other than language, can reveal new insights and knowledge of immense value to a much wider group
of researchers, businesses and citizens, and facilitate a greater number of people working on some challenging
The most urgent climate".
The continued collaboration between IBM and NASA is expected to include the construction of a Foundation Model for weather forecasting
and climate using MERRA2, a dataset of atmospheric observations. This collaboration is seamless
From NASA's Open Science Initiative, which has committed to forming a collaborative scientific community over the next decade.
More of the topic in Hayadan:
