The Value of Machine Learning
How would you make Machine Learning valuable? Since Machine Learning is the competitive assembly of algorithms that improve over time, how do you know what algorithms to assemble? How do you make them improve over time? And can you just “add Machine Learning” to make something better?
Many years ago I worked at a wood mill. We made molding from roughhewn planks that had been sawn from trees. The input was ugly, the output was eighteen or twenty-foot long, perfect, blemish-free lengths of perfect molding.
The lumber’s journey through the mill began with the complete disassembly of the raw planks. Huge boards of pine, oak, or hemlock were chopped up into small pieces — flawless little pieces with no knots, pitch pockets, or wane — that were meticulously reassembled through a finger jointer, pressed, and glued into precise length and dimensions.
At this point, you may be asking yourself “Ok, but what can I learn about Machine Learning from that?”
Nearly all the equipment was standard across every wood mill in America. The difference was down to efficiency — the rate of return from raw board feet in, compared with perfected board feet out. The equipment is the same and the raw material has the same distribution of variable quality. So, what is different, and how can you win?
The answer is the millwright. By the job description, the millwright is required to keep the machinery working. But how the millwright does that work is the difference between profit and loss. Every millwright has the same basic set of tools: wrenches, welders, hammers, and shims. But every wood mill has the same problem: machinery breaks, parts are in short supply, expensive, and with long lead times. Great millwrights know the sounds, vibrations, and feel of the machinery in operation.
They sense when things are wrong before the operators do. And the best millwrights invent their own tools, manufacture custom parts, and are there on the weekends trying new inventions, welding new apparatus to machines, and testing. They know their environment so well that their innovations are pointed, purposeful, and on target.
Simply adding Machine Learning to a process is no different than adding sockets to someone’s toolbox. You’ve seen innumerable Machine Learning toolkits and add-ons in various products and, if you’ve played with them, you’ve ended up deciding it was a waste of time. “How would I use this?” “I’m not using the standard product as it is!” “What does this algorithm do, anyway?”
Lucidum’s Machine Learning and system development teams are millwrights. For the last ten years, our team has been immersed in security operations, incident response, and cyber risk data mining. This immersion led to the careful selection of just the right algorithms which — in concert and in competition with each other — yield the fantastic results Lucidum delivers.
Ten years of security immersion taking standard algorithms, building their own, and welding custom tools resulted in a platform that uses your overlooked data, connects to and empowers your underutilized tools and does it within the first day. The next morning, without understanding algorithms or managing Machine Learning tools, you get complete visibility, recommended action steps, and Lucidum will even go and fix it at your command.
In a wood mill, the difference between success and failure is an immersed and innovative millwright. In your environment, surrounded by unknowns, too many priorities to address, and a flood of new external issues, the difference between success and failure is Lucidum: immersed, innovative, and insightful. See how with a demo.