News

IBM Corp. and NASA have unveiled a new artificial intelligence (AI) model aimed at accelerating the analysis of satellite data. The model is available on Hugging Face, a platform similar to GitHub that allows for the sharing of open-source neural networks. The collaboration between IBM and NASA will continue with a focus on expanding the AI to cover additional use cases, including a partnership with Clark University.

The model has been specifically designed by IBM to help researchers identify areas in the continental U.S. that may be at risk of flooding and wildfires. It boasts the capability to analyze geospatial data up to four times faster than previous state-of-the-art neural networks and requires less data for training.

Described as a foundation model, the AI is based on the popular Transformer architecture, which allows it to take contextual information into account when reasoning about data. This enables more accurate decision-making compared to other AI systems. Transformer models are widely used in advanced AI systems, including OpenAI LP’s GPT-4 large language model.

IBM and NASA trained their model on a geospatial dataset called Harmonized Landsat Sentinel-2, which includes images of Earth’s surface from NASA’s Landsat-8 satellite and measurements from the European Space Agency’s Sentinel-2 satellite constellation. IBM utilized its internally-developed Vela supercomputer, powered by Nvidia Corp.’s A100 series of data center graphics cards, to train the AI model. The supercomputer also incorporates IBM’s virtualization software, which minimizes performance impact.

The model’s potential applications extend beyond flood and wildfire risk detection. IBM and NASA believe it can be adapted for tasks such as deforestation monitoring, carbon emissions tracking, and crop yield forecasting. In collaboration with researchers from NASA and Clark University, IBM plans to further enhance the AI’s capabilities, focusing on time-series segmentation and similarity research.

IBM intends to release a commercial version of the model through its Watsonx product suite, which provides software tools for building and deploying advanced AI models. The suite includes prepackaged neural networks optimized for various use cases. Watsonx is powered by Red Hat OpenShift AI, a component of IBM’s machine learning portfolio specifically designed for AI workloads.

By making satellite data analysis faster and more efficient, IBM and NASA’s advanced AI model will facilitate critical research areas such as climate change and contribute to advancements in various fields.

The post IBM and NASA Release Advanced AI Model for Satellite Data Analysis appeared first on satProviders.