MACHINE LEARNING : THE OPEN SOURCE WAY

DailyPost 772

MACHINE LEARNING : THE OPEN SOURCE WAY

Most of the digital research and product creation and even adoption is mired by company owned source codes / algorithms, patents , intellectual property, transfer of know how and the like. What superficially seems to be a level playing field is meticulously created and managed in favor of the big players. Machine Learning & consequent Artificial Intelligence currently is blessed with open source products based on humungous research & development of lots of MNCs. Whether that would benefit them as part of their long term strategy is a different story, for the time being it is boom time for smaller players to create great application breakthroughs.

Google DeepMind had made it’s machine learning platform public; DeepMind Lab. Facebook did the same with its deep learning software last year, and Elon Musk’s non-profit organization OpenAI released Universe, an open source platform to train AI systems. Some of other frameworks include Apache Singa, Shogun – the oldest in the field, TensorFlow, Scikit-Learn, MLib (Spark), Amazon Machine Learning and Apache Mahout. It can be a field day for the start-ups & entrepreneurs if they really want to stand on their own and deliver. Application is no easy business in ML. Nonetheless, the ecosystem available today couldn’t have been better.

The experts say that a closer look at ML movement and it’s present maturity is because of it being broadly open source always and that open research & development has brought it this far. Thus, open source is the favourable route forward. Open source Machine Learning projects provide some clear advantages; reproducing scientific results & fair comparison of algorithms, quick bug finding & fixing, accelerate scientific development with low cost – reusing methods, long term availability & support and faster adoption of machine learning by various industries.

Still there are no short cuts. The marriage of the domain and technology has to be perfect. Making the machine learn is still not an easy task. It is long, arduous based on a serious undercurrent of human intelligence. AlphaGo and AlphaZero has proved it. If the product / application gets created, it is functional, operational and business magic. It’s undoubtedly worth the effort.

THE ML OPEN SOURCE WAY CAN BE THE FASTEST MODE OF IT’S LARGE SCALE ADOPTION.

Sanjay Sahay

Leave a Comment

Your email address will not be published. Required fields are marked *


The reCAPTCHA verification period has expired. Please reload the page.

Scroll to Top