10 ms·

Sadly it isn't even faster and requires more staff to operate. While the CT pilot program didn’t show reduced wait times at participating airports, the agency

by bhawks 8y ago

Sadly it isn't even faster and requires more staff to operate. While the CT pilot program didn’t show reduced wait times at participating airports, the agency is expecting the technology will speed up lines as passengers and staffers get used to the new processes. Pekoske said he expected more staff to be needed initially as the new machines are put in place.

Someone1234 8y ago
They're planning on using "AI" (Machine learning/Computer Vision) to slowly automate identification. They already do this now, but are limited by the quality of images. > For example, TSA is working to develop new algorithms that use machine learning approaches to discriminate between threats and benign objects, making the screening process more effective and efficient. Machine learning also offers a way to screen for all prohibited items (explosives, firearms, sharp objects, etc.) automatically. It is anticipated that machine learning algorithms not only will improve security effectiveness but also will support automation in future security systems, thereby enhancing operation efficiency and improving passenger experience through increased throughput and decreased false alarm rates. More broadly, machine learning algorithms can be applied to assess security performance and provide system-level improvements beyond performance enhancements realized at individual screening operations. https://www.dhs.gov/sites/default/files/publications/TSA%20-%20Advanced%20Integrated%20Passenger%20Screening%20Technologies.pdf https://www.dhs.gov/sites/default/files/publications/TSA%20-... It is worth noting the limitations of such techniques: Familiarity. Back when Apple switched their Macbooks from HDDs to SSDs, a lot of early purchasers got flagged at TSA because their laptop didn't match a profile (i.e. looking for a spinning stack of platters). With CT technology, we could see even more nuanced differences get flagged for unusual.
- MertsA 8y ago
  >With CT technology, we could see even more nuanced differences get flagged for unusual. Good, that's more data to learn from and that would only be a minor issue for a short period of time for new devices. How often have you flown with something that other passengers hadn't likely have flown numerous times with before you? Even if the system picks out 0.1% of items for manual classification and training, that's still mountains and mountains of data for it to learn from and that would only take a tiny fraction of the time for airport security to identify compared to what they deal with now.
  - Scoundreller 8y ago
    They’ll just add it to your history of items to your permanent record. Just don’t get consistent with luggage and suddenly go unique! The real fun starts when they create a social graph of people carrying similar things. For your safety of course. It probably could feed into an interesting friend-making system if they assigned seats based on the info.
    - MertsA 8y ago
      They could do that now with the current x ray images if they wanted to. Also by classification I'm just referring to classifying in broad categories such as benign, weapon, drugs, bomb, etc. Right now when you go through security the images aren't tied directly to the owner, just a location and a time. Unless the scanner gets tied to e.g. scanning your boarding pass when putting a bin through that doesn't give you a way to tie that information to an individual. Additional cameras with some AI to track people from the boarding pass scan through the checkpoint would allow for what you're talking about but all of that is orthogonal to what specific type of scanner is being used.

by bhawks 8y ago

Someone1234 8y ago

They're planning on using "AI" (Machine learning/Computer Vision) to slowly automate identification. They already do this now, but are limited by the quality of images. > For example, TSA is working to develop new algorithms that use machine learning approaches to discriminate between threats and benign objects, making the screening process more effective and efficient. Machine learning also offers a way to screen for all prohibited items (explosives, firearms, sharp objects, etc.) automatically. It is anticipated that machine learning algorithms not only will improve security effectiveness but also will support automation in future security systems, thereby enhancing operation efficiency and improving passenger experience through increased throughput and decreased false alarm rates. More broadly, machine learning algorithms can be applied to assess security performance and provide system-level improvements beyond performance enhancements realized at individual screening operations. https://www.dhs.gov/sites/default/files/publications/TSA%20-%20Advanced%20Integrated%20Passenger%20Screening%20Technologies.pdf https://www.dhs.gov/sites/default/files/publications/TSA%20-... It is worth noting the limitations of such techniques: Familiarity. Back when Apple switched their Macbooks from HDDs to SSDs, a lot of early purchasers got flagged at TSA because their laptop didn't match a profile (i.e. looking for a spinning stack of platters). With CT technology, we could see even more nuanced differences get flagged for unusual.

MertsA 8y ago
>With CT technology, we could see even more nuanced differences get flagged for unusual. Good, that's more data to learn from and that would only be a minor issue for a short period of time for new devices. How often have you flown with something that other passengers hadn't likely have flown numerous times with before you? Even if the system picks out 0.1% of items for manual classification and training, that's still mountains and mountains of data for it to learn from and that would only take a tiny fraction of the time for airport security to identify compared to what they deal with now.
- Scoundreller 8y ago
  They’ll just add it to your history of items to your permanent record. Just don’t get consistent with luggage and suddenly go unique! The real fun starts when they create a social graph of people carrying similar things. For your safety of course. It probably could feed into an interesting friend-making system if they assigned seats based on the info.
  - MertsA 8y ago
    They could do that now with the current x ray images if they wanted to. Also by classification I'm just referring to classifying in broad categories such as benign, weapon, drugs, bomb, etc. Right now when you go through security the images aren't tied directly to the owner, just a location and a time. Unless the scanner gets tied to e.g. scanning your boarding pass when putting a bin through that doesn't give you a way to tie that information to an individual. Additional cameras with some AI to track people from the boarding pass scan through the checkpoint would allow for what you're talking about but all of that is orthogonal to what specific type of scanner is being used.