Machine Learning: the base line of the future
When people think of machine learning, it calls to mind a range of associations. For some, it’s a deep fear of machines taking over the world, ‘Terminator’ style, or annoying targeted ads by companies that make sinister use of our data as we use social media or browse the internet. Others may picture a utopia where machine learning in our fridge completes our shopping list and sends it to the grocery store while we travel to work in autonomous electric cars without traffic jams or accidents.
The truth is that there are many applications out there that are designed only to make commercial use of our private information and the notion of ‘privacy’ has long disappeared into the depths of the incomprehensible User Agreement. Despite all of this, technically speaking it seems today that the age of truly autonomous cars is right around the corner. Of course, there are many questions of ethics, use of information and responsibility to be addressed before this will become a common reality.
Yet, the field of machine learning is already changing the game in terms of the way we live our lives and conduct our business. According to the website KDNuggets, the top ten use cases for machine learning include anomaly detection, fraud prevention, smart services and recommender systems – the kind of solutions we encounter more and more with every step we take in the digital universe.
Machine learning applications are pushing boundaries in diverse fields such as retail, agriculture and manufacturing, and are even working behind the scenes in places we wouldn’t expect to find them. One such example is the use of machine learning to achieve Fuzzy Matching, which supports integration projects by connecting data sets that are otherwise not connectable, such as unidentified customer records. Other examples are used by information-intensive organizations such as banks and insurance companies, to help classify and catalog data sets. These are just a few internal processes that we, as end users, are not even aware are happening behind the scenes.
The fact is that today, many organizations are investing huge amounts of money in projects for the integration and establishment of data lakes, which require significant mapping and analysis processes. The application of machine learning is already helping to drive down these enormous mapping and development overheads in areas that were, until recently, thought of as being ‘human-exclusive’ and were sometimes an actual roadblock to embarking on these projects, usually due to a high-risk profile or a simple desire to avoid ‘looking under the hood’.
Data usage also entails great responsibility. Today, when GDPR (which for me is more complicated to understand than k-means clustering!) is at the center of attention, the use of data to derive value is becoming harder than ever, forcing companies to consider data anonymization. Machine learning plays a vital role in making this process achievable, with mechanisms that de-identify records, create synthetic data and asses the risk of information disclosure.
The use of machine learning is gaining more and more traction and many organizations (not just Google and Facebook) already have basic and early-stage machine learning applications; some – and not just those that we might expect to – even have highly-advanced applications.
It is interesting to think of a future where our children will not be required to have a driver’s license. But until then, we will see more and more areas where machine learning simply helps human beings to do their job in almost every field, by using smart algorithms.
Today, using these algorithms is not just an option for a company that wants to achieve a competitive edge and become more efficient in areas such as IT, information systems management and all those other applications. It’s an absolute necessity.