Need to survive in a Cambrian AI/SW/HW chaos and accelerate AI research?
Collective Knowledge platform
was developed by researchers to connect multidisiplinary AI/ML/systems communities
and help them share artifacts, workflows and knowledge as
and reusable components
with a unified JSON API and meta information.
assemble novel AI/ML algorithms from such components as LEGO(tm), share them,
and crowdsource their benchmarking, optimization and co-design
across diverse models, data set and platforms from IoT to supercomputers in terms of speed, accuracy, energy and other costs
(see ACM ReQuEST tournaments
All results including mispredictions
are then shared in CK repositories
in a reproducible way
to let the community reuse and improve them!
We hope that CK approach of reusable artifacts and workflows
will help to accelerate adoption of AI while dramatically reducing R&D costs and boosting innovation
(see CK intro slides
We made a special effort to simplify CK installation with minimal dependencies
across Linux, Windows and MacOS. You just need Git client and Python 2.7+ or 3.3+.
Just try it yourself as described here.
You can then follow a simple tutorial to get the first feeling
about CK capabilities (portable, customizable and reusable workflows and artifacts with a portable package manager).
Join the open consortium led by non-profit cTuning foundation and
to participate in the following community activities
powered by the CK.