Machine learning, if you have not already met it, is the capability of a machine (a software application) to modify its rules and algorithms according to new data.
In other words, the machine that learns is one that independently adapts to changes to produce reliable decisions and results now and into the future.
Machine learning is not as new as some other advances in computing, but is coming to the fore, now that technology is available to process very large amounts of varied data (big data) very fast.
So, the question is, is it time to apply machine learning to business continuity as a self-learning watchdog to keep enterprises out of menacing situations?
It can be instructive to look at some of the common uses of machine learning today to see how they might map onto business continuity benefits:
- Fraud detection. One of the more obvious choices for business continuity, especially for financial institutions, the ability to detect, prevent, and mitigate fraud has a direct impact on the ability to stay in business.
- Social media feedback. When machine learning is combined with linguistics, it becomes possible to recognise types of comments on social networks and discover whether people (customers, for instance) are saying good or bad things about you. Given the weight of social media today in buying and boycotting decisions, knowing about potential “bad press” upfront could save a company from sticky times afterwards.
- Recommendation engines. A well-known commercial example is the recommendation system of an online vendor like Amazon (“You might also like…”) that learns from new page visits and purchases of the visitor. The same principles could apply to business environments (“You might also watch out for/protect against…”), learning from changes in markets, customer requirements, and business objectives.
- Self-driving cars. These vehicles are programmed to get from A to B safely, avoid accidents, and handle new situations along the way. Sound familiar? Business continuity plans and management could use the same ideas.
Of course, somebody still has to identify the data needed, organise its collection, write the machine learning algorithms to interpret it, and test the results to make sure they are valid. Any volunteers?