DHS · DHS-135 · record R1914

Low Probability of False Alarm (Low-Pfa) Algorithm for on-person screening.

Department of Homeland Security · TSA

Overview

Development stage: DeployedSource: c) Deployed – The use case is being actively authorized or utilized to support the functions or mission of an agency.
High-impact designation: High-impactSource: a) High-impact
Impact justification: Not reported
Start date: 2022-12-12 (day precision)Source: 2022-12-12T00:00:00
Withheld from public reporting: Not reported

Mission

Topic area: Transportation
Operational functions: No function tag matched Derived by keyword rules; see methodology

Problem the AI is intended to solve

Increase passenger throughput via improve detection performance and decreasing alarm rates and passengers touch rates by 50%

Expected benefits

The purpose is to reduce alarm rates while providing increased passenger throughput and experience. Utilizes Machine Learning (ML) to improve detection performance while decreasing alarm rates and passengers touch rates. The algorithm is gender agnostic which no longer requires officers to select a passengers gender prior to being scanned. Advanced imaging technology (AIT) throughput and AIT utilization have increased with this new algorithm. Note: Once the algorithm is trained, it is locked down and no longer learning.

System outputs

The AI outputs target coordinates to the operator viewing station which is viewed as a bounding box on a representative human figure.

Technology

AI classification: Computer visionSource: Computer Vision: AI that processes and interprets visual data (e.g., images and videos).
System name(s): Not reported
Custom-developed code: Yes
Public source code: Not reported

Sourcing

How it was built: Purchased from vendorSource: a) Purchased from a vendor
Vendor (as reported): Leidos, Rohde & Schwarz
Vendors (standardized): Leidos

A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.

Data and privacy

Involves PII: No
Privacy Impact Assessment: www.dhs.gov ↗
Authorization to Operate: No
Demographic features: Not reported
Training and evaluation data: Vendor AITs are tested in a laboratory environment using mock passengers. Statistical tests are done for false and true alarms, a performance measure for detection. Statistical tests are done for probability of false alarm and probability of detection, a performance measure for detection capability.
Federal Data Catalog: Not reported

Governance

8 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.

Pre-deployment testingReported in placeSource: a) Yes

AI impact assessmentReported in placeSource: a) Yes

Independent reviewReported in placeSource: c) Yes – by the CAIO

Ongoing monitoringReported in placeSource: a) Yes, sufficient monitoring protocols have been established

Operator trainingReported in placeSource: a) Yes, sufficient and periodic training has been established

Fail-safeReported in placeSource: a) Yes

Appeal processReported in placeSource: a) Yes, an appropriate appeal process has been established

User and public consultationReported in placeSource: a) Direct usability testing; b) General solicitations of feedback and comments from the public

Potential impacts and how they were identified

Security risks include false negatives allowing threats to get throught the sterile side of the airport or high false alarm rates slowing opperations.