About Machine Learning
Machine Learning (ML) is one of the most fascinating technologies I’ve ever come across. It breaks the notion most of us software engineers grew up with professionally which is that, to solve a business problem, you start with the process/rules (algorithm), and the data, and then write a program to apply those rules to the data and come up with the answer.
ML breaks that sacred paradigm and turns the equation around completely. It takes the data, and the expected answer as inputs, and then it comes up with the rules. It mimics the way we humans learn. For example, when we grew up as infants, our parents taught us what a banana looks like, perhaps once, perhaps a few times. Now we’re able to recognise a banana immediately, regardless of its position and location. What “rules” do our brains apply to recognise such object? What neuronal connections were created in or brains to allow us to identify a banana in milliseconds? And where exactly are those rules stored in our brain? Nobody knows for sure (yet) but the fact remains, we are able to remarkably and accurately identify bananas, amongst thousands of other objects by just looking at them. A ML model works exactly the same way; you feed it with enough pictures of bananas (training) and it will learn to recognise them (inference) on its own by implementing a set of rules (model).
Applications of Machine Learning to solve business problems
And ML is indeed impressive. We have seen ML applied to recognise people in your social media photos, partly understand the hand-written text from a doctor, accurately predict rainfall, and recommend what you should watch next on TV based on you’re into. More recently, ML Large Language Models have taken us by surprise with their accuracy and ability to hold a “human” conversation.
However, how many times have you seen ML actually solving a business problem, automate a process, and deliver evidenced value in the form of cost reduction, a better client/employee experience, or reduce operational errors? I argue that these examples are fewer and farther between.
And so, I want to share a couple of examples from our recent experience working with organisations, where ML has been deployed to solve an actual business problem. In these posts, I’ll share the problem we solved, the technologies we used (spoiler alert, it’s never a single, almighty product), and, more critically, the business outcomes achieved with the solution as well as a few of the lessons we learned along the way to hopefully help you as you embark on your own ML projects.
I’ll also do my best to equip you with a few simple questions that would be tough for a non-practitioner to answer regarding applied ML so you can get behind the PowerPoint slides as evaluate solution providers.
Please stay tuned for these and get in touch for a more in-depth discussion.