Federated Machine Learning Testbed

ARCC Miscellaneous Services

ARCC is pleased to offer Federated Machine Learning as an upcoming service. Federated Learning is a relatively new decentralized approach to training large machine learning and AI models. This is a framework for collaboratively training an AI model across differential privacy and homomorphic encryption with multi-party computation.

Find out more about how Federated Learning can be a viable solution for large scale AI model training.
Read the 1 Page Whitepaper

Navigation Main Page

ARCC Services

LEARN MORE High Performance Computing Data Storage Services User Support Specialized Services

Federated Learning to Address Data Privacy & Scarcity

Over Capacity!

Training AI models requires very large datasets and large scale, unified cluster hardware compromised of high capacity storage and powerful compute nodes. With the growth of AI and AI training datasets, requirements can exceed the capacity available at a single research facility.

Exceeding capacity at a single Location means data and computations must be distributed among multiple facilities ⇒

Security Concerns!

Sharing of data between multiple decentralized locations opens large training datasets with sensitive information to data possible breeches as several copies of the data are shared, transferred or stored among multiple locations.

Data security must be maintained, and Federated AI offers an answer ⇒

Federated AI as a Solution

Federated Learning relies on edge computing and distributed computing frameworks where processing of raw data takes place on local devices close to where it's originally generated. Only the results from local processing of data get sent to external sources to train the model in collaboration. Most federated learning consists of training local AI models, then transmitting encrypted weights of the models to an external authority to compute the global model.

Read the 1 Page Paper ⇒

What is Federated AI?

Typically, Artificial intelligence (AI) models require extremely large volumes of data and previously, the large-scale datasets had to be centralized in a single location when training a model. This requires either massive unified infrastructure that is available to host and compute the data in it's entirety for training, and creates a vulnerability by opening up opportunities for any personally identifiable information (PII) contained in the datasets to be exposed any time data was subject to transmission or storage.

Federated learning addresses these concerns. Sensitive information remains on the node, preserving data privacy. It also allows for collaborative learning, with varied devices or servers contributing to the refinement of AI models.

1-PG Whitepaper On Federated Learning

Requesting this service

To get answers to specific questions

E-mail us below with any inquiries, questions or to learn more:

E-mail ARCC-HELP@uwYO.EDU

To make a request for a new federated AI project

Fill out a request form on our ARCC Service Portal

Our Portal

Learn more about Federated AI

To learn more about federated learning and AI

Read about Federated Learning

Contact Us

Advanced Research Computing Center

Office of Research & Economic Development

1000 E. University Ave
Laramie, WY 82071
Dept. 3355
Email: arcc-help@uwyo.edu
ARCC Technical Wiki

The University of Wyoming has earned its Research Level 1 (R1) status from the Carnegie Classification of Institutions of Higher Education, placing Wyoming's only four-year university with the top research universities in the United States.