A new machine learning solution developed by Montreal-based AxiPolymer Inc. is designed to make it easier for large-sized plastics processors to embrace artificial intelligence.
September 27, 2018 by Canadian Plastics
The manufacturing world has probably seen more new technologies in the past few years than Wile E. Coyote unleashed on the Road Runner in total, but none are bigger than the advances in artificial intelligence (AI).
Defined as human intelligence exhibited by machines, the general idea behind AI in an industrial setting is the ability for machines to think for us, whether they’re projecting sales forecasts or directing a factory assembly line. And a new machine learning solution developed by Montreal-based AxiPolymer Inc. is designed to make it easier for large-sized plastics processors to embrace AI. Founded last year by polymer scientist Dr. Ata Zad, AxiPolymer provides product development services for the plastics sector. And no product deserves development more than AI, Zad said. “Every plastics manufacturer has the potential to integrate machine learning into their operations and become more competitive by gaining predictive insights into production,” he said. “Machine learning’s core technologies align well with the complex problems manufacturers face daily, especially large manufacturers that have the largest amount of raw data.”
But adaptation of AI by the plastics sector has been slow. “In many manufacturing companies, IT systems aren’t well developed and aren’t aligned together to act as a total decision aid system, which makes it difficult to bring all of the different parts of a company onto the same path to accomplish shared goals,” Zad said. And without AI, the problem is only going to get worse. “When a company expands, the amount of new data it has to deal with grows exponentially — data relating to new clients, new formulations and products, new suppliers, new employees, and more — and it’s almost impossible to interpret these huge loads of data and get actionable intelligence from it with a traditional IT system,” Zad said. “The only way to do this is with AI, which uses algorithms to find hidden patterns in data. These algorithms are iterative, and will learn continually and seek optimized outcomes; they also iterate in milliseconds, which lets manufacturers find optimized outcomes in minutes instead of months.”
GOING DEEP THROUGH DEEP FEATURES
Now available from AxiPolymer after more than six months in development, the machine learning system is both modular and highly customized. The first step towards implementation, Zad said, is for AxiPolymer to meet with a company’s executives and find out what their business priorities are, and where exactly they need help from AI. Determining a company’s so-called “deep features” is key. “Deep features are features that can only be obtained with the help of field experts, as opposed to shallow features that can be obtained directly from observations,” Zad said. “Once we understand the deep features, we start gathering the data from the company that needs to be analyzed. We then feed that data into an algorithm that will begin to get trained and start predicting patterns, some of which will be right and some wrong — the system will teach itself and learn enough to become operational after a short period of time. At that point, the customer can run the AI system on their own.”
And since each plastics processor is unique, each machine learning solution will also be unique. “AI isn’t a one-size-fits-all technology,” Zad said. “It’s company-specific and policy-specific, which means every company needs its own customized deep features and customized AI algorithms, and we deliver that. We also introduce the AI solution in a modular format — to the supply chain first, for example, and then to production, and then to sales and marketing. We finish by connecting them all together.”
When it comes to the benefits of using AI, Zad said, the sky’s the limit. “AI can bring new levels of insight, intelligence, and predictive accuracy to every phase of plastics production,” he said. “It can increase production capacity, optimize production workflows, help achieve a more accurate cash flow, improve preventative maintenance, create more accurate sales forecasts, improve supplier selection to make sure you source from the right suppliers, power analytics for more effective advertising, and more,” he said.
AI might even have allowed Wile E. Coyote to finally bag that pesky Road Runner.