An advanced artificial intelligence (AI) approach developed by researchers at Los Alamos National Laboratory offers a method for reconstructing extensive data fields, like ocean temperatures, using minimal field-deployable sensors.
This low-powered Edge computing technique has far-reaching applications across various sectors including industry, science, and medicine.
Javier Santos, a Los Alamos National Laboratory researcher, has been instrumental in developing a neural network that efficiently represents large systems in a compact format. “We developed a neural network that allows us to represent a large system in a very compact way,” Santos explains. “That compactness means it requires fewer computing resources compared to state-of-the-art convolutional neural network architectures, making it well-suited to field deployment on drones, sensor arrays, and other Edge-computing applications.”
The Senseiver: a novel AI technique
The team’s work, published in Nature Machine Intelligence, introduces a novel AI technique called Senseiver. Building on Google’s Perceiver IO model and applying techniques from natural-language models like ChatGPT, Senseiver reconstructs broad-area information from limited measurements. Dan O’Malley, a co-author, and Los Alamos researcher, highlights the model’s efficiency: “Using fewer parameters and less memory requires fewer central processing unit cycles on the computer, so it runs faster on smaller computers.”
This model has been validated with real-world sparse data and complex three-dimensional-fluids datasets. In a significant application, the team used the Senseiver to process sea-surface-temperature data from the National Oceanic and Atmospheric Administration. This data, collected over decades from satellites and ship sensors, was used by the model to forecast ocean-wide temperatures, contributing valuable insights to global climate models.
Los Alamos National Laboratory finds the Senseiver model particularly useful. Hari Viswanathan, a Los Alamos National Laboratory Fellow and co-author, notes: “Los Alamos has a wide range of remote sensing capabilities, but it’s not easy to use AI because models are too big and don’t fit on devices in the field, which leads us to Edge computing.” This model is expected to benefit drones, sensor networks, and other applications currently limited by advanced AI technology.
The AI model shows promise for a variety of practical applications, including self-driving cars, remote asset modelling in oil and gas, patient medical monitoring, Cloud gaming, content delivery, and contaminant tracing. In particular, the model will be instrumental in Los Alamos National Laboratory’s work in identifying and characterising orphaned wells under the Department of Energy funded CATALOG programme.
