Written by gjw

Artificial Intelligence is behind many of today’s technological advances. Today cars can drive themselves safely, robots can see and interact with our shared environment, autonomous agents can intelligently support us to live better, and teaching agents that “know everything” can support us as we learn more. No agent is intelligent without learning and the development of many algorithms for machine learning support the growing intelligence of our autonomous agents. And this technology comes to the fore in supporting the exploration of the masses of data we collect today through data science, analysing more data that has been stored in the recent past than the total of all data every, prior stored.

The machine learning hub (known between us who know as the MLHub) is an exciting new platform that brings the ongoing amazing advances in AI, machine learning, and data science into the hands of anyone interested to exploring and building on this technology. Through the MLHub we openly and freely share AI agents, pre-built Machine Learning models, and and the very best of Data Science practices. The growing spread of package hosted on the MLHub is also augmented by new developments exposed on GitHub MLHub wraps these exciting, yet often incredibly complex and difficult to use functionalities into a command line and graphical tool that is easy to get into. But beyond ease of use, the MLHub packages make it possible for anyone to build on this technology to bring your own ideas and your own visions to to explore, to share, and to contribute to the good of humanity. 

MLHub links git based repositories into a collection of quickly accessible and ready to run, explore, rebuild, and even deploy, pre-built machine learning models and data science technology.  The models and technology are accessed and managed using the ml command from the mlhub package. The package is available for quick installation from pypi. A growing number of machine learning models and data science technology are becoming available, as well as cloud based services.

If you have access to a computer running Ubuntu 18.04 LTS (e.g., Windows 10’s Subsystem for Linux) then getting started is simple. Install the command line tool to explore the rain, colorize, objects, and facedetect machine learning models, to start with. Have a look also at the data science visualisation packages, including animate and beeswarm. Explore the offerings from the Azure cloud like azcv and azspeech2txt But why not explore each one of them! There may be examples there that really interest you or could support your current projects and mobile applications.

MLHub works best on Ubuntu 18.04, ideally on your own laptop. It is also really easy to install on Windows 10 through the Windows Subsystem for Linux or the Hyper-V gallery (enable Hyper-V and choose Ubuntu), and MacOS X (using Parallels or Virtual Box to install from the Ubuntu iso). It also runs well on an Azure Ubuntu server or a Ubuntu instance on any cloud server.

The models on MLHub are curated, but any model on github can support the MLHub framework. To have a model curated please send mailto:mlhub@togaware.com. It is easy to contribute your own model simply by creating a configuration for it called MLHUB.yaml and a demo.py or demo.R file on your github repository. Then simply point the ml command to that repository!