By Anabel Cooper – Blogger,
Machines and Humans Team up for Cyber Security
When it comes to computing high volumes of mathematical calculations, there’s no way a human can compete with a computer. In terms of both quantity and quality of accurate data processing, machines have humans licked. The role they play in our daily lives is becoming more and more voluble by the day, as we look forward to a tomorrow populated by connected cars, voice activation and facial recognition software. However, there’s no way machines are ever going to be able to replace humans when it comes to performing certain roles. Computers themselves are one such area where human guidance is still crucial when it comes to directing their applications and pointing them in the right direction of relevant data. And, of course, the human brain itself fires more calculations in a second than the most powerful computer ever could. Nonetheless, in the field of cyber security, combining the two will be essential to the future of digital security.
Machine learning is a subsidiary field of artificial intelligence that aims to provide machines with sufficient data for them to learn new processes themselves and develop response functions under their own initiative. Today, machine learning predominantly manifests itself in developing prediction models to help with forecasting. Its applications within cyber security are even more tantalising. Cyber security staff are forever engaged in a constant battle with criminals targeting governments, individuals and companies for financial or political gain. If it takes a cyber security analyst an average of, say, fifteen minutes to investigate and resolve a single security alert that means that an individual can only handle around thirty alerts a day.
New Threats Emerge
That simply doesn’t constitute enough of a defense to deal with the hundreds of thousands of security threats that the biggest financial institutions have to deal with on a daily basis. Across every industry, a workforce shortage is emerging as companies struggle to staff themselves with sufficient numbers of cyber security personnel from a labour pool that is struggling to supply companies with the cyber security human resources that are crucial to navigating the world of modern cyber crime. Criminals online are only too aware of this emerging vulnerability and are becoming emboldened by what they perceive to be the dawn of a new golden age for cyber crime. One of the problems is that many weapons in the cyber criminal’s arsenal are incredibly virulent, able to replicate themselves from device to device, while each of their incursions has to be resolved on an individual basis.
A New Line of Defense
To that end, the cyber security industry is looking ever more closely at the possible applications of machine learning in helping to combat the myriad threats lurking online. The automations of scoping, advanced classification and prioritisation of security events, collectively known as Analytics 3.0, is helping an ever increasing number of CSOs get the most out of security and human resources working in tandem to respond to data with enhanced accuracy and better response times. It’s helping IT staff find and analyse vulnerabilities and collate the relevant data so that it’s ready and waiting when the time comes to act on it effectively.
Best of Both Worlds
The concept of machine learning is decades-old, but now computers exist with the processing power necessary to implement learning algorithms on a wide scale. When properly implemented, the result is that calculations delivered by machine learning can power endpoint security processes like remediation and cleanup before malware has even had a chance to execute. Working together, human-machine teams can tighten endpoint security without detriment to the user experience or a big increase in resource expenditure. Big business have been the fastest to implement machine automation within their cyber security assets, albeit with the knowledge that human agency is needed at every stage to help guide these systems in the right direction. There are burgeoning developments in this field on the horizon, as machine learning processes could eventually be applied to automate the detection and analysis of new and emerging forms of cyber attacks.
One such example of innovations in this field emerged at Defense Advanced Research Projects Agency (DARPA) and their Computers and Humans Exploring Software Security (CHESS) initiative. Its conception began at DEF CON, the world’s foremost hacker and cyber security convention. One of the regular events was a competition for spotting security vulnerabilities. While the competition was ostensibly open for humans, Shellphish, a team of computer science majors from UC Santa Barbara decided to enter their software into the competition instead, demonstrating the tantalising possibilities automated learning could have in cyber defense. As business and cyber security communities begin to realise the potential of man-machine cooperation, we can look forward to seeing many more innovative examples of these collaborations a lot sooner than we might think.