Harnessing machine learning for baggage scans -- GCN

The Transportation Security Administration is looking to Silicon Valley startups to help it bring machine learning to security screening to improve the accuracy of airport baggage scanners. Through an Other Transaction Solicitation, the Department of Homeland Security's Science & Technology Directorate and TSA's Office of Requirements and Capabilities Analysis are looking for a new way to detect evolving threats carried in airline passenger luggage. Rapidly changing consumer electronics, the RFI said, are an example of a dynamic threat vector that evolves faster than next-generation detector hardware. TSA personnel looking at baggage scanner images might miss subtle new differences in how newly introduced consumer devices are wired or put together. The agency wants developers to come up with AI-based methods that could automate detection algorithm training, allowing detection hardware to "intuitively recognize" such subtleties and new objects that come through airports in luggage.