Federated Learning Using In-Space Data (FLUID )

Overview

Development stage

In development or acquisitionSource: a) Pre-deployment – The use case is in a development or acquisition status.

High-impact designation

Not high-impactSource: c) Not high-impact

Impact justification

Not reported

Start date

Not reported

Withheld from public reporting

NoSource: a) No

Mission

Topic area

Science

Operational functions

No function tag matched Derived by keyword rules; see methodology

Problem the AI is intended to solve

Enables neural net models to be trained using a combination of terrestrial data and space-borne data without the need to downlink or uplink data for consolidation and training in the normal manner. Instead, small components of the overall neural net model are trained in-situ with the data and transmitted for federation into a single neural net, thereby reducing data transmission demands and reducing overall latencies by several orders of magnitude.

Expected benefits

Enables neural net models to be trained using a combination of terrestrial data and space-borne data without the need to downlink or uplink data for consolidation and training in the normal manner. Instead, small components of the overall neural net model are trained in-situ with the data and transmitted for federation into a single neural net, thereby reducing data transmission demands and reducing overall latencies by several orders of magnitude.

System outputs

For example, a neural net model to monitor lunar habitat astronauts for signs of lung toxicity due to regolith inhalation could be largely trained on Earth using similar data (e.g. data collected for volcanic ash inhaltion) and then fine-tuned with data generated in-situ on the lunar surface, without the need to transmit that data back to Earth. The FLUID arhitecture has been fully tested using the Spaceborne Computer-2 on board the ISS; in February of 2024, the world's first neural net model trained with both terrestrial and in-space data was successfully trained.

Technology

AI classification

Classical / predictive MLSource: Classical/Predictive Machine Learning: Models trained on data to make predictions or classifications based on identified patterns or relatio…

System name(s)

Not reported

Custom-developed code

Not reported

Public source code

Not reported

Sourcing

How it was built

Not reported

Vendor (as reported)

Not reported

Vendors (standardized)

None on the standard list

A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.

Data and privacy

Involves PII

Not reported

Privacy Impact Assessment

Not reported

Authorization to Operate

Not reported

Demographic features

Not reported

Training and evaluation data

Not reported

Federal Data Catalog

Not reported

Governance

0 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.

Potential impacts and how they were identified

Not reported