This Project Grant from the National Science Foundation's Division of Computing and Communication Foundations, under the Computer and Information Science and Engineering federal grant program (CFDA 47.070), provides $199,999 to support the development of novel algorithms and schemes for ensuring robust operations of deep neural networks. The award to Texas A&M Engineering Experiment Station, doing business as Tees, will fund the Guardiann project from March 1, 2023 through February 28, 2026.
Guardiann has two components. First, it will develop dynamic monitoring schemes using neuron pattern-based and model-based algorithms to determine whether a deep neural network model could misbehave during inference in a production environment. Second, it will create testing schemes to explore how models might behave in corner case scenarios and allow users to examine model behavior without understanding internal model details. Together these monitoring and testing approaches aim to provide trust in deep neural network operations.
Generated 1/7/24, 12:12 PM