Taking Stock: The Internet of Things, and Machine Learning Algorithms at War

It’s in the news every day; hackers targeting banks, hospitals, or, as we’ve come to fear the most, elections.

Suffice to say then that cybersecurity has, in the last few years, gone from a relatively obscure industry – let’s qualify that: not in the sense of importance, but rather how folks have been interacting with it – to one at the forefront of global efforts to protect our data and applications.

 A decade ago, cybersecurity researchers were almost as caliginous as the hackers they were trying to defend folks against, and despite the lack of fanfare, some people still chose it as a career (*gasp).

We spoke to one of our whizz-kids, Gilad Yehudai, to find out what makes him tick and why, of all the possible fields in tech, he chose cybersecurity at a time when it might not have been the sexiest of industries.

Protecting data and applications, a different beast altogether

One of the major challenges facing the industry is the ability to attract new talent; especially when competing against companies that occupy the public sphere from the moment our alarm wakes us up to the moment we lay our phones to rest. Gilad, who has a master’s degree in mathematics and forms part of our team in Israel, offers a pretty interesting perspective,

“The world of cybersecurity is a fascinating one from my point of view, especially when trying to solve machine learning problems related to it. Cybersecurity is adversarial in nature, where hackers try to understand security mechanisms and how to bypass them. Developing algorithms in such environments is much more challenging than algorithms where the data doesn’t try to fool you.”

Never a dull moment

Additionally, our industry is one in flux, as more threats and vulnerabilities are introduced, and hackers find new ways to bypass security mechanisms. The latter was a pretty big draw for Gilad, whose experience in mathematics and serving in the Israeli Army’s cyber defense department made him a great candidate for the Imperva threat research team.

“The research group at Imperva seemed like the perfect fit, as large parts of my day to day job is to develop machine learning algorithms in the domain of cybersecurity, and the data I use is mostly attacks on web applications.”

Speaking of attacks, Gilad and the rest of our research team sure have their hands full.

“In my opinion, the Internet of Things (IoT) security is one of the biggest challenges out there. More and more devices are connected to the internet every day and these devices may be put to malicious use. Hackers may enlist these devices to their botnet in order to launch attacks like DDoS, ATO (account takeover), comment spam and much more.”

Worse still, our growing network of ‘micro-computers’ (smartphones, tablets etc.) could be manipulated and their computational power used to mine cryptocurrencies.

“Protecting these devices the same way we protect endpoint PCs will be one of the biggest challenges.”

Change brings new challenges, and opportunities

On the topic of change, the cybersecurity industry, according to Gilad, is headed increasingly towards machine learning and automation; which serves us well.

“If in the past most security mechanisms were based on hard-coded rules written by security experts, today more and more products are based on rules that are created automatically using artificial intelligent algorithms. These mechanisms can be much more dynamic and adapt better to the ever-changing world of cybersecurity.”

That said, the more the industry relies on machine learning algorithms for defense, the higher the likelihood that hackers will look to manipulate those same algorithms for their own purposes.

“Hackers may try to create adversarial examples to fool machine learning algorithms. Securing algorithms will require more effort, effort that will intensify as these algorithms are used in more sensitive processes. For example, facial recognition algorithms that authorize access to a specific location may be fooled by hackers using an adversarial example in order to gain access to an unauthorized location.”

While the cyber threat landscape continues to evolve, and the bad actors looking to nick our data and compromise our applications get increasingly creative, it’s good to know that there are experts whose sole purpose it is to ‘fight the good fight’, so to speak.

“Research is a bit like walking in the dark, you don’t know in which direction to go next, and you never know what you are going find. Sometimes you begin to research in some direction, and in the process you find a completely other direction which you haven’t even though about at the beginning. Research is not for everybody, but I get really excited about it.

*** This is a Security Bloggers Network syndicated blog from Blog | Imperva authored by Gerhard Jacobs. Read the original post at: https://www.imperva.com/blog/2018/09/the-internet-of-things-and-machine-learning-algorithms-at-war/