Machine Learning

Machine learning is a special branch of artificial intelligence.  Sadly, a lot of the terminology used in discussion of these subjects is ambiguous and designed to suggest equivalence with humans at some level.  At this point, none of the applications of artificial intelligence are anything of the sort.

There are, however, a number of applications in which the speed of the computer is used to solve problems by doing simple tasks that are programmed tasks millions of times per second in order to accomplish some greater task.  We generally refer to this as the ‘brute force’ approach to problem solving.  For example, if we talk about discovering a 4 digit password, we know that a computer can try all possible numerical combinations in a matter of seconds.  If there is an interlock after the third try that shut out the user, or in this case, the hacker, then the brute force method will not work.

Writing a program to try every numerical combination for a 4 digit password is a human activity that creates instructions for a binary computer to execute a programmed solution.  Writing a program that searches for correlations between a group of variables and reporting their values is another set of instructions for a binary computer to execute.  We can do research all the types of statistical relationships between data and come up with computer code that executes those instructions and produces results.

This is not intelligence.  It is the process by which a binary computer mimics human intelligence.  It takes a lot more intelligence to create the binary instruction mechanism, nowadays a semiconductor processor.

We are seeing the beginnings of a fundamental shift in how we do things in the manufacturing environment.  Maybe the “AI” tools really represent machines that supervise control systems.  We need some new terminology to deal with the ambiguity of how the systems are deployed and where things sit in the physical world.

Notwithstanding the complexity, the driving force is value.  The new tools are being created because there is money to be made.  AI in the automation world means the ability to apply smarter tools to the everyday operations of manufacturing companies everywhere.  As the cost for these tools decrease, become more widespread with demonstrated benefits, they will be indispensable.  Higher productivity and lower downtime are values that many large business will pay for.

Maybe we are seeing the age of the Jetsons coming into view.




Mechatronic Tips