FedML is a decentralized, open and collaborative machine learning platform that facilitates training, deployment and continuous improvement of AI models. The firm has recently announced its collaboration with Konica Minolta, a Japanese multinational firm that manufactures business, industrial and medical imaging devices. FedML operates under the notion that the future of AI is large scale collaboration. The platform aims to make AI collaboration open by combining data models and computing resources on cloud infrastructure to enable training, deployment and monitoring of AI models to take place from anywhere. The FedML AI platform aims to offer MLaaS(Machine Learning as a Service) for AIoT(Artificial Intelligence of Things), FinTech(Financial Technology), Healthcare, BioTech, Precision medicine, Autonomous driving, Manufacturing, Generative AI, Gas and Oil development and Energy among many other industries. The platform also has a token economy to enhance AI collaboration on data models by availing compute resources.
FedML has recently announced that it has raised $6 million to facilitate collaborative AI by empowering companies and developers to work together by sharing data, models and compute resources in the process of fine-tuning ML(Machine Learning) models. The firm aims to fuel AI innovation beyond the technology giants that exercise monopoly. To meet these obligations, FedML has developed an open-source community enterprise platform complete with a set of tools that make it easier for developers to train and deploy ML models at a very huge scale across cloud nodes everywhere. The ecosystem created by FedMl is aimed at helping businesses and enterprises customize and deploy AI models to bring efficiency in customer services, business automation, content creation and product designs. Because data regarding companies and industries is sensitive and heavily regulated, traditional cloud-based AI training solutions are not suitable fro this task. It is only decentralization coupled with AI that can solve this problem with security, transparency and efficiency.
In its operations, FedML has discovered that federated learning is the most efficient solution for companies that use private datasets to train AI models at the edge without need to share their sensitive data to other companies, a phenomenon it has dabbed ‘learning without sharing’. This will enable firms to customize their own models without pulling customers private data. These scenario can be applicable in the medical field where health companies can use scarce datasets to develop AI models for rare disease detection without exposing sensitive patient data. This forms part of the basis why FedML has formed partnerships with Theta Networks and Konica Minolta to supply such applications.
For developers, FedML has an open-source library that offers access to a unified and scalable distributed computing framework from anywhere at any scale. The open-source community on FedML ecosystem also offers state-of-the-art models that are compatible with its platform. Software developers only require to install the library on their computers using a CLI(Command Line Interface) tool and from there they can invite collaborators to create new AI groups that can also serve as a network of collaborators on the ecosystem. The developer features on the platform are immense as users are allowed the ability to select an aggregation server and even select edge devices.
