TechRepublic's Dan Patterson asks Schneider Electric Chief Digital Officer Herve Coureil about how machine learning will transform industries. The following is an edited transcript of the interview.
Dan Patterson: What feeder industries? I mean, it's one thing for us to kind of nerd-out about your particular sector, but there are ancillary industries. What companies, what industries, do you see really taking advantage of machine learning?
Herve Coureil: We are working. In the segment we serve, we see, for instance, in water, in oil and gas, which are really industries that are very intensive in their use of energy, where really, downtime is super costly. We see them as very much in the forefront of IoT. And then you hear also a lot about the term industry 4.2, which is really the manufacturing processes, right, where, again, you want to instrument your manufacturing processes for the sake of efficiency and ability.
But I would say, yeah, speaking specifically for AI, for instance, one of the early application that we had of AI were actually in oil fields, in rod pumps. Instrumenting the rod pump, and having actually an AI algorithm that was machine learning-based, that actually would provide predictive maintenance on those pumps. And that was actually pretty interesting from a couple of perspectives.
One is that it's actually a good use case of transfer learning, because each pump has slightly different operating conditions, right, so you want to be able to have your AI, or your machine learning model, be adaptive.
The second thing, which was actually really interesting, is again, at the convergence of IoT and AI, is the fact that this is really machine learning that you train on the cloud, but then you infer on the edge, because you don't want the user in the remote field, so you don't want to have network dependency, you don't want to have a round trip with the cloud. You want to have in real time, et cetera.
So, actually, many of those industries that I was just mentioning are really industries that will soon have the forefront of applying machine learning, but not necessarily as a full cloud thing, but really also pushing the inference as far on the edge as you can. Because again, in those critical processes, you don't want to have network dependency, you don't want to have... you want to be as deterministically real-time, as you can.
- What enterprises will focus on for digital transformation in 2018 (ZDNet)
- 10 tech tools that will help bring digital transformation to your SMB (TechRepublic)
- Digital transformation: Three ways to get it right in your business (ZDNet)
- 8 digital transformation resolutions for CIOs in 2018 (TechRepublic)
- The 6 rules of digital transformation: How IT can get in the game (TechRepublic)
- How digital transformation is reshaping the IT budget: The journey of 3 CIOs (ZDNet)
Dan Patterson has nothing to disclose. He does not hold investments in the technology companies he covers.
Dan is a Senior Writer for TechRepublic. He covers cybersecurity and the intersection of technology, politics and government.