Dr Michael Cheng: finding new ways to detect online fraud
Cheng was not always an online detective. “Before postgraduate studies, I worked at ARUP, an international civil engineering consulting firm, for about a year,” Cheng said, “Given a relatively stable geology in Hong Kong, civil engineering projects mostly involve the effects of static forces on objects at equilibrium. While this is an important technique and is sufficient for most construction projects in Hong Kong, what really interested me is the study of dynamic behaviours and how the strength, stiffness, and stability of structures are impacted by changing forces such as earthquakes. Mathematically this is a much more complex problem and is what fascinates me”. After seeking some advice, and with support from the Croucher Foundation, Cheng started his doctoral studies with Professor Thomas Heaton, a world-renowned seismologist and the Director of the Earthquake Engineering Research Laboratory at Caltech.
Cheng’s main project at Caltech was to develop a low-cost and scalable community seismic network. This network could be used to alert local residents seconds before an earthquake – a short but valuable time for people to run to safety – or to trigger automated actions in buildings, for example opening the doors of fire stations. Detailed “shake maps” could also help emergency services to prioritise rescue efforts.
At a cost of tens of thousands of dollars each, and requiring expert installation, existing seismometers would be impractical in a community seismic network which might need to cover millions of homes in a metropolitan area. Working with a team of geologists and computer scientists, Cheng used accelerometers to develop a small inexpensive seismometer to collect ground motion intensity data before, during, and after earthquakes and wrote custom software to analyse data feeds from thousands of devices. Data from the Community Seismic Network is used during an earthquake to provide very high resolution data on actual ground shaking to emergency services. In the longer term, the data will enable scientific construction of 3D geological models of the ground underneath the sensors, which can be used to influence land use policy and construction regulations.
After completing his PhD, Cheng decided to join Signifyd which was, at that time, a small startup with less than 10 employees. “I was the first data scientist with a machine learning background to join the company”, Cheng recalled, “I worked with other engineers to develop the first fraud detection model and to build a production version which could screen millions of transactions in real time.” Signifyd now employs over 100 employees and, in addition to playing a leading role in technology development, Cheng also serves as a direct point of contact within the company for several very large online retailers.
The WannaCry ransomware attack in May 2017 brought global attention on data security. As a global financial centre, Hong Kong was warned at “serious risk” by Hong Kong Applied Science and Technology Research Institute. Police statistics showed that financial losses from cyber crime surged by 26% last year, amounting HK$ 2301 million of lost.
For e-commerce out there who are concerned of data security, Cheng gave the following advice. “As the engagement in the internet grows in our daily life, it is getting harder to keep our personal data secure. People share their personal information frequently to retailers or on social networks, and the data is either public or is prone to data breach. That is why companies like Signifyd exist to protect the e-commerce industry and help to eliminate the worry by shoppers and merchants.”
Dr Michael Ming-Hei Cheng is currently a Senior Data Scientist at Signifyd, USA. He obtained his bachelor’s degree in Civil Engineering from The University of Hong Kong with first class honors in 2008. Under the support of Croucher Scholarship and the George W. Housner Earthquake Engineering Research Fellowship, he completed a PhD in Mechanical and Civil Engineering at the California Institute of Technology, USA, in 2014.
To view Dr Cheng’s Croucher profile, please click here.