Transfer Learning will power Knowledge Transfer

In humans AND machines.

Machine Learning and DNN’s will be at the Center

In a sense this is both scary and a new reality. We need to be ready. Because like most things that seem likely to happen … this is one that will happen whether we are ready or not.

Partners and Platform Providers can play a big role

I see this as an opportunity for platform providers like Microsoft, Google, IBM, SAP and others. I also see this as an opportunity for partners. Where partners that work independently and also with these and other platform providers can create solutions that either stand alone of plug into these platform providers partner ecosystems.

Microsoft Research is getting ahead of the curve.

imageHumans are Retiring & Humans are Changing

Two of the facts humans will be dealing with in the next few years are related to Baby Boomers and Millennials:

  • 10,000 Baby Boomers are reaching retirement age
  • 10,000 Millennials are reaching the age of 21

imageThere needs to be some serious Knowledge Transfer from the retiring Boomers. However, that transfer may NOT be to another human. It may be to a robot.

Partners and Platform Providers that build solutions to make this as easy as possible have an opportunity. Of course, there are laws, logistics and costs to consider. The fact is … there is an opportunity here.

Not to mention the fact that as automation takes over there will be fewer “traditional” jobs for humans. This gives rise to an uptick in free time and a possible need for a Basic Income. But, that’s another topic for another post.

imageThis post was inspired by a tweet from Microsoft EVP in AI and Research, Harry Shum. This is a topic I have long explored and have worked on dozens of projects, but all of them were focused on the human element.

This is a different take and one I had not been thinking about. The Transfer Learning capabilities of machines. It is 100% logical and it makes sense. It will be.

Knowledge Transfer and Transfer Learning

SNAGHTML157fb845Imagine the worker on the production line. Whether it was an old school production like manufacturing automobiles or a more modern version making iPhone’s.

  • How do they transfer their skills to a new employee?
  • Whether they are moving up, leaving the company or the company is growing.
  • They need to have a way to transfer their skills.
  • How do they do it?
  • How have you done it? How will a robot do it?

This has been a human effort for years. Things were documented in 3 ring binders and on collaborative app platform (e.g. SharePoint and other ECM tools) – tasks, schedules, timelines, people to got to in a pinch, in an emergency, etc.

Guess what? This process is coming to a robot / machine near you!

Whether it's a robot or just a set of chips (perhaps a gelatinous goo) it's going to be the way machines learn too.

Machine Learning
The scientific study of algorithms whose capabilities improve with experience.
Source: Peter Lee, Microsoft CVP of Research on Edge

Boomers Moving On & Millennials Moving In

Remember: Every day … 10,000 Baby Boomers are reaching retirement age and 10,000 Millennials are reaching the age of 21.

How will we deal with Transfer Learning?

Microsoft CVP of Research Peter Lee has written a great article on the concept and realities of Transfer Learning.

Some of these advances in Deep Neural Networks, or DNN's, have led to computing systems that are competent in skills that are associated with human intelligence, sometimes to levels that not only approach man’s capabilities but, in some cases, exceeds it.
Via Peter Lee on Edge (emphasis is mine)

Robots are Not Perfect

Robots are often suited for a specific task. One in which they excel. If there is such a thing for a robot. But, the fact is … Transfer Learning combined with Machine Learning combined with Artificial Intelligence is making it more practical and possible to allow a robot (a machine) to learn as they go.

I use the term “robot” here as a catch all that includes machines that may not have a physical form. For example, IBM’s Watson is a collection of programs. It may use Voice Detection combined with image processing and other input sources to learn.

Robots and machines continue to get smarter every day. I don’t envision a Terminator like future, but I do expect to see a lot of “traditional jobs” replaced by automation. Not necessarily by a physical robot, but that will happen too. Rather, the jobs that can be automated will be automated.

There is a heated discussion going on around the world today about the use, abuse and perhaps mis-use of automation. Some of the political leanings today seem to imply a potential Luddite-like uprising. Perhaps this will happen in some areas, but ever since the Agrarian Age led to the Industrial Age there have been jobs replaced by automation and jobs created by automation. I’m an optimist. I think it will balance out and make the human condition better.

The idea that Transfer Learning will power some of these efforts is still in it’s early stages as noted in Peter Lee’s article.

“Transfer learning opens the possibility that all the intelligence of the web can form the foundation of machine-learned systems …”
Via Peter Lee on Edge (emphasis is mine)

Although it’s early days these are the times where partners with a vision can take on some of this early thinking and see what they can build with it. I suspect the partner programs from Microsoft. IBM, and even from Saleforce type companies will begin to offer platform components that allow partners to start building today.

We’ve seen this with Blockchain and Big Data. It’s time for Transfer Learning to enter the mix.

What do you think?

  • Are you ready for some Transfer Learning?
  • Will you train a robot? Or will they just watch you and learn?
  • Will you be trained by a robot? Will they get frustrated and just do it themselves?

Add your thoughts here. Or, ask your robot to do it for you.

Image Credits: 4erevolution

clip_image001Jeff is business advisor, mentor and community engagement expert. He has spent most of his career in the Enterprise Content Management industry. He brings over 20 years of Channel Sales, Partner Marketing and Alliance expertise to audiences around the world in speaking engagements and via his writing. He has worked for Microsoft, Kodak, and K2.

Connect with me on Twitter @jshuey

Or connect on LinkedIn, Facebook, or Google+

I am a a contributing author to Entrepreneur, Elite Daily, Yahoo, US News and to the Personal Branding Blog



Lee Andrews said…
This was informative. Btw, Do you know about Binfer? Very easy tool to transfer big data.