It’s the time of the year where the world is bombarded with predictions for the following year. But rather than talking about various predictions, I’ll focus on the one that we believe will have a broad long-term effect:
The use of machine learning (ML) will grow rapidly across all aspects of cybersecurity. With recent advances in technology using machine learning is no longer a dream, or confined to certain aspects such as academics. Rather, ML is fast becoming stereotypical, with open-source and commercial offerings targeted toward cyber defense. Massive compute and storage capacity at affordable prices and infrastructure-as-a-service (IaaS) and platform-as-a-service (PaaS) offerings are making Machine Learning solutions more affordable and easier to install.
More importantly, the algorithms that revolve around machine learning focused around supervised and unsupervised Machine learning and the toolkits, are rapidly advancing in capability and maturity, particularly in the finance sector. With the availability of user-friendly toolkits, APIs and third-party integrations, massive parallel graphics processing unit (GPU) systems in Machine Learning-based computation is now becoming mainstream. With these advances, threat intelligence researchers and technology companies hope to use ML-based solutions as a way to tackle cyber threats and build more accurate baselines into normal behavior and surface anomalies against that.
Key to the success of this is access to relevant and targeted at training data that is used for supervised and unsupervised Machine Learning. Here, too, the content-rich network traffic data and metadata, which were extremely hard to come by, are now becoming easily available. This is one promising and long-term trend that can significantly advance cyber defense.