security

Security driven by AI and machine learning – Sports Business Journal


Ashim Banerjee chuckled when asked about IDmission’s initial interest in the sports venue space. That’s because there was none, the IDmission CEO said. “We had no idea that the space even needed us,” Banerjee told SBJ.

That was until Aramark reached out in the last couple of years to help with the age/ID verification problem. IDmission started in 2011, originating out of the idea to establish identity securely through smartphones, an industry that was starting to attract more consumers.

Slotterback said that Aramark learned of IDmission through another partnership. And that initial outreach began with the simple question of “can you do this?” As previously mentioned, Aramark had looked at and experimented with other self-service options — Slotterback pointed out that earlier this summer, Aramark worked with Amazon One contactless pay for self-authentication and payment at Coors Field as an example. That effort, though, still required an interaction with a person. Slotterback pointed out that IDmission and Zippin were both willing to spend a ton of development time toward integration and making the tech work for the guest. “We landed on IDmission because of their ability to be flexible and nimble with us,” Slotterback said.

IDmission has integrated with Cecilia.ai and Drink Command in the self-service space (and previously with TendedBar). It also worked with Wicket on the Browns’ express beer-authentication program.

Katie Beattie, the director of marketing for Aramark’s design and development team, added that the initial phase of a pilot process features many “nitty gritty” conversations, especially when working on a project that’s made from scratch. Once the time came for beta testing, Beattie said the team reviewed numerous demos and screenshots, going through the signup process as well to get a firsthand look at user experience. “Essentially, for lack of better words, we tried to break it,” Beattie said. “Can we break this? The answer was no.”

What helped set IDmission apart even more was the security of the profile and transaction. This works due to IDmission’s passive liveness detection. The system can determine the difference between a human face and an attempt to deceive (like a photo of a person, an image of someone’s face on a screen or a person wearing a mask). Banerjee said this function took IDmission a year-and-a-half to create with the help of AI and machine learning. He said IDmission used roughly 1.5 million images — half of them of live people and the other half featuring deception attempts — as part of the training. IDmission even took the tech to Madame Tussauds in Las Vegas to test on the lifelike wax figures, which the AI could differentiate from flesh and blood.

Banerjee said IDmission has to retrain a new model every two weeks to enhance the AI’s capability, a process that’s created at least 60 new models. “It’s a very tedious, long, drawn-out affair,” Banerjee said. “But once you do it, it works.”

With a quick face scan, users can enter a Zippin store, where they can shop and leave quickly, without having to take out their wallet or phoneAramark





READ SOURCE

This website uses cookies. By continuing to use this site, you accept our use of cookies.