To develop the ecosystem on all aspects in the context of Machine Learning, from native compilation to high-level abstractions and auxiliary projects.
Integrate with state-of-the-art tools for native numerical compilation. The goal is to target different accelerators (CPU/GPU) and with different modes (AOT/JIT)
Provide high-level abstractions for different disciplines of machine learning, such as neural networks, supervised learning, clustering, etc
Develop auxiliary projects that are essential to numerical computing and machine learning efforts, such as interactive tools, plotting libraries, etc
Our goal is to enable the Erlang Ecosystem to be used for Machine Learning and other numerical computing tasks. We believe functional programming can be a good fit for numerical computing, especially as we move to higher-level abstractions.
The Machine Learning WG also plans to collaborate with other Working Groups on many efforts. For example, Build and Packaging can help us discuss how to ship precompiled libraries. The Machine Learning and Embedded WG can work together on to provide tooling that runs on the edge. The External Process Communication can assist us to integrate with native tools.
This working group can benefit from the foundation in several ways:
The foundation can provide us communication tools so the Machine Learning WG becomes the main hub for ML discussions in the ecosystem
The Foundation can facilitate collaboration with interested WGs and relevant efforts happening in the ecosystem
The Foundation can fund some of the tasks outlined in our long-term deliverables