Author Bio ▼

IFSEC Global is the online community for the Security and Fire industry. Our market-leading live events span the globe, connecting buyers and sellers.
January 13, 2017

Sign up to free email newsletters


Wireless Access Control Report 2021

Applying deep learning to cyber security: Q&A with Deep Instinct CEO Guy Caspi

guy caspiDeep Instinct is the first company to apply deep learning to cyber security.

Guy Caspi, the Israel-founded company’s CEO, spoke to about the complexities of deep learning and how Deep Instinct spotted a gap in the ballooning market for combating cyber security threats.

This interview was originally published on Before discussing today’s cyber security threat environment and Deep Instinct solutions in greater detail, please tell us about your background and company history?

Guy Caspi: I’ve utilised my advanced degrees in Mathematics, Machine Learning and Business to apply mathematics and machine learning in a technology elite unit of the Israel Defense Forces (IDF), as well as in financial institutions and intelligence organizations around the world.

Over the past two decades, I’ve led some of the largest government cyber and big data projects in Israel and other countries.

Founded in 2014, and out of stealth mode since November 2015, Deep Instinct is the first company to apply deep learning to cyber security. With offices in Tel Aviv, Israel and in North America, we now have 65 employees.

Our company has a winning combination of people who have the academic knowledge and credentials, paired with unique experiences in cyber security gained through years in the intelligence and elite units that focused on cyber- attacks. We adopt the mindset of hackers in order to be prepared for all vulnerabilities.

In addition, Deep Instinct’s dedicated deep learning research group is headed by one of the leading researchers in the field of computational intelligence. Moreover, the company has a highly-experienced management team that leverages its cyber security and academic backgrounds to carry out a successful product that offers an effective solution to address a critical need in the industry. Your site claims that Deep Instinct is “The first company to apply deep learning to cyber security”. Please give us an overview of how Deep Instinct works.

GC: Deep Instinct’s core technology is deep learning, which is an advanced branch of artificial intelligence (AI). Deep learning is inspired by the brain’s ability to learn: once a brain learns to identify an object, its identification becomes second nature.

Similarly, as Deep Instinct’s artificial brain learns to detect any type of cyber threat, its prediction capabilities become instinctive. Deep learning has exhibited groundbreaking results when applied to computer vision, speech, and text understanding and we are the first company to apply it to the cyber security domain.

In cyber security, there a big need for solutions that can protect against brand new (zero-day) threats in real-time – a critical issue that causes great vulnerabilities to almost every business.

Deep learning is complex and its application has a very high barrier entry because the neural networks are comprised of tens of hundreds of layers and the mathematics required to create such layers is extremely difficult.

Even once this hurdle is passed, the implementation of running massive data sets using GPUs (Graphic User Interface) is not an easy feat.

Moreover, creating a deep learning-based technology that can run as an on-device client requires great expertise that raises the bar even higher. The few companies that have demonstrated these capabilities have mostly been acquired by giants, such as Google, Facebook and Salesforce.

Furthermore, Deep Instinct does not use open source deep learning libraries but instead, has created its own. What are the major benefits with respect to detection, prevention, accuracy, ease of deployment and other features.

Guy Caspi: Deep Instinct offers a unique solution of prevention, which includes blocking malware before it is activated and can cause harm. Many new solutions on the market can only offer detection and prevention once the business has been infected, but we can detect and prevent before any damage occurs.

By way of analogy, if a business were a person and the malicious attack were poison, other cyber security vendors need the person to first touch the poisonous object to then they can act and prevent the poison from spreading throughout the body. From Deep Instinct’s perspective, we can tell the person not to touch the poisonous object in the first place because we immediately identified it as harmful.

Moreover, Deep Instinct focuses on unknown threats and APT (Advanced Persistent Threat) attacks whether they are in a file-less manner or already existing in the system. Instead of waiting for the next unexpected attack, the next unknown attack is identified and blocked in real-time before any harm can occur.

The ability to offer immediate prevention extends beyond a network or Internet connection by covering the device even when it is not connected to them.

Furthermore, our detection rates are substantially higher that existing solutions on the market. This unprecedented accuracy in predicting unknown cyber threats is enabled by the application of proprietary deep learning algorithms.

Deep learning’s capabilities of identifying malware from any data source results in comprehensive protection on any device, platform, and operating system, filling in gaps by providing complete solutions. Finally, deployment is fast and seamless and the solution’s operations do not affect the user experience.

Click here to read the full interview on


Get your summer security fix in this essential free 'State of the Nation' webinar

Explore the state of security in the United Kingdom in this unmissable webinar led by industry titans Professor Dave Sloggett, Surveillance Camera Commissioner Tony Porter, TSI's Rick Mounfield, BSIA's Mike Reddington and Alex Carmichael of the SSAIB.


Related Topics

Notify of
Inline Feedbacks
View all comments