A competition hosted on the data science community site, Kaggle, seeks to improve the Department of Homeland Security's threat recognition algorithms.
The US Department of Homeland Security (DHS) wants the help of data scientists to boost the algorithms its uses to detect airport security threats, and it's offering $1.5 million in prize money for the best solutions. The competition was created in partnership with data science community Kaggle, which is owned by Google, and aims to lower false alarm rates in security screenings.
The security screening process at US airports isn't typically regarded as the most efficient or pleasant part of air travel. Unfortunately, it also isn't very secure. In 2015, it was found that the Transportation Security Administration (TSA) failed 95% of security tests conducted by DHS. While processes have likely improved since then, the addition of machine learning and neural network-powered tools could help make the system even more secure.
SEE: Deep Learning: Convolutional Neural Networks in Python (TechRepublic Academy)
The biggest bottlenecks for travelers in the security line come when the scanning equipment algorithm registers a potential threat, and a TSA agent must manually screen the traveler, the competition page said. The goal of this competition is to reduce the number of false alarms that require that additional screening.
According to the Kaggle competition page, the TSA sources its algorithms directly from the manufactures of the scanning equipment. These manufacturers have slow release cycles and can be expensive, so it's now looking to the data science community for new options.
"Using a dataset of images collected on the latest generation of scanners, participants are challenged to identify the presence of simulated threats under a variety of object types, clothing types, and body types," the challenge post said. "Even a modest decrease in false alarms will help TSA significantly improve the passenger experience while maintaining high levels of security."
The competitors will be providing with a sample set of body scans from a TSA machine. Each body will be divided into 17 standard zones. For a given scan, competitors must predict the probability that a threat is present in each zone, the competition page said.
Of the $1.5 million offered, $500,000 will go to the first place winner. Second place will get $300,000, while third place will receive $200,000. The fourth through eighth place winners will each be given $100,000.
Deadlines for entry into the contest start on December 4, 2017.
The 3 big takeaways for TechRepublic readers
- The DHS is hosting a competition on Kaggle to improve the algorithms used in its security screening process at US airports.
- Data scientists will need to predict the probability of a threat in 17 body zones on each scan, in order to lower the risk of false alarms.
- A total of $1.5 million in prize money will be available to the winners of the competition, and deadlines begin on December 4, 2017.
- Understanding the differences between AI, machine learning, and deep learning (TechRepublic)
- Google's DeepMind claims major milestone in making machines talk like humans (ZDNet)
- Google weaves AI and machine learning into core products at I/O 2017 (TechRepublic)
- Google is using neural networks to turn your selfies into cartoon stickers (ZDNet)
- Using AI-enhanced malware, researchers disrupt algorithms used in antimalware (TechRepublic)