Restricted Research - Award List, Note/Discussion Page
Fiscal Year: 2023
2090 The University of Texas at Dallas (143978)
Principal Investigator: Chen,Feng
Total Amount of Contract, Award, or Gift (Annual before 2011): $ 837,007
Exceeds $250,000 (Is it flagged?): Yes
Start and End Dates: 4/13/23 - 5/1/23
Restricted Research: YES
Academic Discipline: Computer Science
Department, Center, School, or Institute: ECS
Title of Contract, Award, or Gift: Hidden Activity Signal and Trajectory Anomaly Characterization
Name of Granting or Contracting Agency/Entity:
Kitware, Inc
CFDA: 16
Program Title: IARPA
Note:
DoD Flow through; [TA-2] Develop uncertainty-aware graph neural network single anomaly detection: Develop an uncertainty-aware multi-scale graph neural network (UM-GNN) to detect anomalous single movements as out-of-distribution (OOD) ones at different scales, in comparison with in-distribution movement classes, such as background single movements and recurrent movements. We will detect OOD movements based on vacuity, a novel uncertainty measure proposed in our prior work (Neurips 2020), that can effectively quantify predictive uncertainty of a GNN model due to lack of evidence.
Discussion: No discussion notes