Cracking down on crime with AI
Computers are being taught to decipher wrongdoing at football matches and more
As our society grows more digitally sophisticated, so too do cyber criminals.
Cyber security has become an increasingly important issue in the 21st century as hackers target governments, businesses and individuals all over the world.
Attacks from hackers are virtually constant, with one every 39 seconds, according to a study by the University of Maryland. And the damage they cause is huge. A report from global security advisory specialists the Herjavec Group predicts that cyber crime will cost the world $6 trillion (£5.2 trillion) annually by 2021.
Recent examples include the hack of hotel chain Marriott International in 2018, during which data were stolen from up to 383 million customers, including passport and credit card information. In 2017, an attack on Qatar News Agency highlighted the risk cyber crime poses to national security.
But, even as hackers develop more complex and cunning ways to launch attacks, others are fighting back.
Researchers at Qatar University are using AI to expose the criminals lurking in the virtual shadows.
Somaya Al-maadeed, head of the computer science and engineering department at Qatar University, has received more than $5 million in grants for research related to image processing and artificial intelligence. One of the latest projects she led was one uncovering criminals’ hidden online messages.
“The name of the project is ‘prediction of covert communication and untrusted channels’. These criminals and hackers hide their communication in content and it is being used to protect communication among criminal members,” says Professor Al-maadeed, whose expertise includes image processing and facial and handwriting recognition.
“Criminals can communicate via messages hidden within YouTube for example, or Facebook or another app,” she explains. “They can hide messages, documents and data in an image or video. So, it is not easy to detect and it is a very serious problem.”
The researchers have developed a system – the first stage of which has a patent pending – to use AI and machine learning to detect these criminal communications. Using existing examples, a trainer “teaches” a computer what to look for, to the point where it can predict which content has hidden messages embedded within it.
“We need good data to train the system and the more images we have, the more accurate the system will be,” says Professor Al-maadeed.
“AI is learning by collecting data from both structured and unstructured sources. With this system, we can eliminate process errors that occur with humans, and use machine learning and deep learning techniques to predict cybersecurity risks.”
Qatar’s Ministry of Interior and a private company have expressed interest in the project, one of several studies focusing on AI at Qatar University.
Professor Al-maadeed has also secured funding to lead studies on crowd modelling and facial recognition in preparation for the FIFA World Cup, to be hosted by Qatar in 2022. The crowd modelling element aims to develop, for the first time, a fully automated system to manage football crowds. This will include technology to show people where to enter and exit a stadium and quickly target fans who need first aid.
The facial recognition study is about training a system to identify international hooligans and alerting organisers when they come to a stadium.
A competition run by Qatar University also invited entrants to find ways that AI innovations could help solve challenges during the World Cup, in areas such as referees’ decision-making and detecting social media trolls.
“The fourth industrial revolution is based on automation and AI. So, we need to be at the forefront in AI research at Qatar University and in Qatar in general,” says Professor Al-maadeed.
“We have strong research groups here at the university working on artificial intelligence and we are teaching courses related to artificial intelligence to our undergraduate and graduate students. And not only those in computer science and engineering: we are proposing a course in artificial intelligence that any university student can take.”
That commitment to an area of research of growing global importance was reflected in the 2020 IEEE International Conference on Informatics, IoT, and Enabling Technologies, hosted by the department of computer science and engineering in February.
The conference brought together international experts to share ideas and advances in areas including artificial intelligence, informatics, computer vision and the Internet of Things.
As AI becomes incorporated into our daily lives, there are challenges that will have to be confronted.
“Some people, for example some within older generations, do not understand and trust AI,” says Professor Al-maadeed. “But based on our experience, if we have enough data, AI can give better results in many fields than humans.”
To tackle the challenges associated with AI and continue advancing the technology, academic freedom is essential.
“We have full freedom to exhibit our research and we encourage students to do so,” says Professor Al-maadeed.
Such is the potential of AI that Professor Al-maadeed is also involved with projects to help detect cancer and teach children Arabic.
“We use a special microscope and infrared to detect colon cancer and then we have our own machine learning techniques to teach the machine how to detect cancers. Colon cancer cells also look similar to breast cancer cells, so we have implemented it in breast cancer detection as well,” she says.
“I’m also co-principal investigator on a teaching project. I’m working with a university in the US and another in Canada to develop an AI system to help teachers teach their students to read and write and also how to pronounce words in Arabic.”
Partnerships and collaborations both within Qatar and internationally enable Qatar University to exchange knowledge with government and industry. As well as working with the Supreme Committee for Delivery & Legacy on innovations for the World Cup 2022, other AI research collaborators include the Qatar Ministry of Transport and Communications, Hamad Medical Corporation and Qatar Petroleum.
Professor Al-maadeed also works with colleagues from international universities, including Northumbria University in the UK, to investigate the latest trends in cyber security and forensics.
She believes that collaborations like these are vital for researchers to keep pace with a global threat that is constantly evolving.
“It’s crucial because we need to transfer and share knowledge from well-established universities around the world. This is especially important when we are working on the latest advances in technology to tackle cybercrime.”
For more information, please visit www.qu.edu.qa