IBM Systems Lab Services

Building a community around machine learning and AI

Share this post:

Community matters to machine learning

Let’s be honest. Machine learning is a complicated space! Two minutes on Google looking into the topic is guaranteed to leave you more confused than when you started. So where can you find help from others who are working in this space?

If you’ve ever thought “AI sounds pretty cool; we should probably be doing that,” you’ve no doubt also asked yourself:

  • Where to get started
  • Which framework to choose (there are probably millions)
  • Whether to use a cloud or physical infrastructure
  • Where to deploy your model
  • How to collate or model your data

There are numerous questions to ask and design choices to make before you even start doing machine learning, and most don’t have one right answer. But there are also numerous resources online to help inform your decisions.

In every corner of the globe, people are researching new machine learning techniques, new ways to model data and derive better insights or find new routes to market. In the time it’s taken me to write this blog post, someone will have written an article on a new, better way of defining neural networks.

At first glance, this wealth of information may seem daunting, but it’s nothing to be afraid of; it’s a jewel in the crown of the AI/machine learning space. The fact that there are so many people building tutorials, creating demos and writing articles and blogs is awesome.

This community of collaborators, who often don’t even realize they’re collaborating, is the reason machine learning and AI are being adopted so quickly and the technology is improving at the rate it is. We wouldn’t be where we are today without that community.

Open source frameworks and the OpenPOWER Foundation

Take TensorFlow, for example. Google decided to make its machine learning framework and tooling open source because sharing these amazing technologies with a wider community helps them grow and helps mature the platform to solve issues for other organizations.

tensorflow computations, Machine Learning Community

There are 1,325 contributors to the TensorFlow project working together to build the best software tooling for machine learning. Without this community, we wouldn’t be using machine learning as readily as we can, and we certainly wouldn’t have the tutorials and articles that make this space accessible.

And it’s not just TensorFlow. There are hundreds of frameworks striving to make machine learning accessible through better tooling and better documentation.

In fact, it’s not just software!

Machine learning brings a unique set of infrastructure challenges. It’s a hugely complex technical compute challenge that puts a massive strain on physical systems. The OpenPOWER Foundation is doing the “open source” thing for hardware — applying the principles of open collaboration and innovation to the servers themselves.

We’re now designing and building systems specifically engineered to tackle these workloads as quickly as possible, meaning that we can train our models faster and build bigger things!

Meetups and real people

Everything I’ve mentioned so far involves digital interactions between collaborators, but finding a community around machine learning doesn’t have to happen digitally. You can get off the internet (once you’ve finished reading and sharing my blog post of course) and go meet people in your area who are coming together to share their passion, knowledge and challenges with AI and machine learning.

IBM Systems Lab Services has several PowerAI meetups specifically focused on accelerated frameworks and IBM Power Systems for you to get involved in across the world.

There’s a whole wealth of machine learning Meetup communities too. Through these events you can meet people to discuss the technology and how you can use it in your business. In my experience, everyone in these groups comes with the same open, altruistic mind-set that infuses the digital communities. It’s about sharing ideas and learning from each other. It’s about meeting face to face, getting to know people who are willing to help each other and share expertise.

So how can you get involved with this incredible movement?

  • Get on meetup.com and find your local PowerAI/TensorFlow/machine learning meetup.
  • Share your ideas, expertise and passion.
  • Comment on blog posts (like this one).
  • Reach out to people on Twitter and ask questions.

That’s it — it really is that easy!

Please get involved, as I know the machine learning community would love to have you — wherever you are and whatever your skill level or interest. Find your local PowerAI meetup and if there isn’t one in your area but you’d still like to be involved, contact us about starting one.

IBM Systems Lab Services Consultant, IBM Power Systems and Machine Learning

More IBM Systems Lab Services stories

Is your scientific software complemented by the best hardware?

Did you know you could realize huge performance gains for your scientific applications by taking advantage of the unique capabilities of the IBM Power Systems platform? Are you apprehensive about making the leap to another architecture? Let’s look at the current state of the software ecosystem around IBM POWER, the resources available and how IBM […]

Continue reading

Top IBM Power Systems myths: Migrating to Power from x86 is costly, painful and risky

“Migrating your Oracle databases to IBM Power Systems is dangerous, expensive and very high risk.” That advice came from a well-known consulting company to a client planning to migrate its mission-critical databases from x86 to Power Systems. Such comments are not unusual, and it makes no difference if the migration is to AIX or Linux. […]

Continue reading

Optimizing data lake infrastructure

In today’s world, data is the new oil, and there’s a great need to preserve that data for exploration and to derive value. A “data lake” acts as a repository that consolidates an organization’s data into a governed and well-managed environment that supports both analytics and production workloads. It embraces multiple data platforms, such as […]

Continue reading