AI coming to the aid of security-related applications

 - 
03/20/2019

Our May 2019 News Poll got me really thinking about Artificial Intelligence (AI) and Machine Learning (ML), and the possibilities. My previous AI-related thoughts have been around Watson, the IBM-created, question-answering computer system that answers in natural language, and robots, and how AI can take over the world one day, according to some! Spooky! But, I wanted to know if AI is a legit, practical application for security-related functions, so I scoured the internet and found some exciting and unique, currently deployed uses. 

Physical Security

According the to China Morning Post, AI is revolutionizing physical security in Asia. It can detect people acting out of the ordinary and flag them, and then transmit that information to a command center, where human operators can make an informed decision. Additionally, AI and high-definition cameras can work together to first communicate to a human that a smoke detector, for example, has been activated, with the cameras identifying the exact location of the fire. 

Financial Security

Shoplifting literally costs billions of dollars here in the United States, which trickles down to honest consumers who end up paying more for goods and services. Vaak, a Tokyo-based company, spent more than 100 hours showing their AI system closed-circuit television footage of honest shoppers and shoplifters. The system can now identify suspicious activity based on more than 100 aspects of shoppers’ behavior including gait, hand movements, facial expressions, clothing choices and even “restless” and “sneaking” behaviors. Store employees are alerted of suspiciousness via an app and they can decide what to do. 

Life Security

Paris-based startup, Pharnext, was founded by Daniel Cohen, who “mapped” the human genome and demonstrated it is possible to use Big Data and automation to speed up the processing of DNA samples. Today, Cohen is using AI to analyze and map the chain of reactions of disease in the body. With this information, he and his team are combining existing drugs, known as “repurposing,” to create therapeutic effects that each drug lacks on its own. His overall goal is to use existing medicines to treat all disease, preventing the design of new medicines. 

Cybersecurity

Post-doctoral research fellow at Stanford University, Dr. Srijan Kuman, is developing an AI method — REV2 — to identify online conflict using data and machine learning to predict internet trolling before it happens. (Trolling is an action by a person who posts inflammatory and often deceptive and disinformation online to provoke others to respond on pure emotion.) Kuman uses statistical analysis, graph mining, embedding and deep learning to determine normal and malicious behaviors. His method is currently being used by Flipkart, an online store, to identify fake reviews and reviewers, and he was able to accurately predict when one Reddit community will troll another. 

Be sure to check out our editor’s blog that talks about worldwide spending on AI systems to reach $35.8 billion in 2019, according to International Data Corporation.