In recent years, data science has emerged into the limelight and gained more attentive scrutiny by business leaders, academics and practitioners. Many have opined that this new field has become a driving force in technology markets, generating tremendous innovation and growth. At its core, data science is an interdisciplinary field that employs automated methods to analyze massive amounts of data, incorporating concepts from different disciplines such as: computer science, mathematics, statistics, analytics and modelling.  

Data science is also a driving force behind BioCatch’s technology. With a multidisciplinary team of engineers and PhD’s, BioCatch’s data science team conducts quantitative research to support R&D operations, reveal hidden patterns/predictive analytics, and using machine learning algorithms to transform large amounts of data into a usable product. The team is also highly service-oriented, routinely communicating with BioCatch’s customers, providing periodic updates on fraud trends and receiving the user’s input.

Dr. Shira Mintz, Head of Analytics at BioCatch, shares her thoughts on how data science is driving the industry forward and what makes BioCatch stand out in the realm of behavioral biometrics.


Q: Tell us a little about your academic background, your transition to the industry and your role at BioCatch?

A: I completed my PhD in Bioinformatics at the Weizmann Institute of Science (Rehovot, Israel). More specifically, I implemented a model building algorithm and reconstructed the first tissue-specific plant cell metabolic model. This model was applied to generate in-silico predictions of metabolic engineering strategies for enrichment of different plant compounds.

During my PhD studies, I managed independent research projects, initiated collaborations with multidiscipline laboratories, both computational and experimental. The computational work included developing tools for large-scale analysis of different ‘omics’ data, including genomics, proteomics and metabolomics. These types of analysis gave me the skills for handling copious amounts of data, and revealing the hidden information within large and noisy datasets.

After I completed my PhD, the transition to the industry was quite fast. An interesting position in a large company dealing with fraud prevention opened up. To be honest, even though the position sounded extremely interesting, I had many doubts since it was not related to biology – my real passion at the time. Nevertheless, I decided not to pass on the opportunity, and I’m very happy about that decision.


Q: How is the extensive use of data science driving the industry forward?

A: In the hi-tech industry, there is no doubt that the need for data science teams is already well established in almost all areas, both in start-ups and corporations. Stating the obvious, as we collect more data, we need more skills and resources to extract the patterns and behaviors hidden within the data. In this process we use huge datasets, join them with other sources, clean them from disturbing noise, and finally come to conclusions which are then used to build the product. Moreover, and perhaps less obvious, in the dynamic environment of the high-tech world, as a company progresses, the challenges keep on changing and the data never stops flowing. Our main goal in this environment is to have the ability to continuously analyze the changing data, suggesting new business directions.


Q: From a data scientist's perspective, what are the main challenges your clients in the banking sector are facing?

A: When people think about the banking sector, progress and innovation are not the first words that would come up… However, working with many banks I can share the feeling that the banking industry has evolved greatly in the last few years, both in terms of operations and service delivery. While doing so, the data collected by this industry has also increased significantly.

Everyone knows that one of the biggest problems faced by the banking sector is stopping or at least reducing fraud. But that’s not all. Seems like the banks are in a constant conflict between their responsibility for the security of their clients’ accounts, while keeping their service delivery uninterrupted and frictionless. Lowering the friction is important for the banks not only in order to make sure user experience is optimal, but also to reduce extensive operational costs. Without the use of big-data, these contradicting demands are extremely hard to achieve in parallel. The use of big-data allows pinpointing the riskiest activities, while keeping the actual end-users uninterrupted. It also allows the bank to achieve sustainable work resources, losing less money in the effort of reducing fraud. 


Q: Can you tell us about an interesting fraud case that your team uncovered and how you were able to discover it?

A: Social Engineering is one of the biggest challenges that banks are facing nowadays. We are often asked by our customers for assistance as this kind of fraud is quite hard to detect. One type of social engineering, called Vishing, is when the fraudster pretends to be a bank representative, calling a customer which he knows all the details about, and convinces him to take action that may not be in their best interest, to say the least.

Interesting cases that are occurring more and more, are when fraudsters convince the victim to enter his/her credentials, login to the system, and then through persuasion and manipulation, to cause the customer to willingly hand over the session through remote access. The fraudster then quickly performs transactions, and before you know it, the money is gone. 

Through in-depth research on different cases, we noticed that the mouse and keyboard activity collected during the session can help us detect remote activity that was performed by the fraudster. What is even more interesting is that we could also spot that the remote activity did not occur during the login stages. Combined with other behavioral patterns seen in the post-login activity we could make sure that suspicious sessions scored extremely high, allowing the customer to block the transaction.


Q: From a data scientist's perspective, what makes BioCatch's solution superior in today's marketplace?

A: A lot has been said about behavioral analysis as the new direction for fighting fraud. The behavioral approach has evolved not just as another method to reduce fraud, but has actually arose from severe shortcomings of other methods currently used in the marketplace. 

Over the last few years, BioCatch has become a key player in the world of behavioral biometrics fraud prevention. Changes in how customers access their accounts or shop through the web required changes in how businesses prevent fraud, and this is where BioCatch comes into play. So what is so unique about our solution? Why is it working so well? In one sentence – it’s extremely difficult to change who you are…every person behaves uniquely, whether they realize it or not. This page is too short for me to list the amount of different behavioral patterns users exhibit each time they log into an online-banking site, and we at BioCatch collect them all. In fact, we record the behaviors of millions of people around the world, and create profiles for each and every one based on these patterns. A user who produces behavioral patterns that differ from his/her own profile is suspicious. Moreover, we don’t only profile the legitimate users, but we also profile the actual fraudsters. This allows us not only to authenticate the user, but also to match behaviors to known fraudulent profiles, ultimately, mitigating risk even more.

As simple as it sounds, it is actually quite difficult. Starting with the massive collection of data – just imagine that every single move you make with your mouse is stored, and ultimately must result with a risk score.  This requires deep understanding of the big data world, extensive research and use of sophisticated automatic tools and techniques. Using these techniques, BioCatch is able to catch a huge portion of current fraud, and is also able to prevent the fraud before it even occurs. 


Q: Where do you think behavioral biometrics is going in the next five years?

A: Behavioral biometrics arose as an emerging alternative to traditional protection methods which are simply not enough anymore. By now it is already well-established that passwords are far less than optimal, either because they can easily be stolen, cracked, copied or because we simply don’t remember them, having to replace them often, causing a high drop-off rate.

Behavioral biometrics has so many advantages compared to alternative methods. Exact human behaviors are extremely difficult to mimic as they are comprised of numerous behaviors; It authenticates the user at all points of time without him/her even noticing it; and it can even determine non-human behavior - all are very important for our evolving industry. These advantages are likely to be employed in more and more sectors around the world - online retailers/e-commerce, banks, smart phones, smart watches, enterprise security and more.



*** Shira Mintz, Ph.D is Head of Analytics at BioCatch. She holds a Ph.D in Bioinformatics from the Weizmann Institute of Science - Rehovot, Israel



Related Posts