Certified AI via Software Analysis and Verification

Project Member(s): Sui, Y.

Funding or Partner Organisation: Commonwealth Scientific and Industrial Research Organisation (Data61) (Data61 CSIRO)
Commonwealth Scientific and Industrial Research Organisation (Data61) (Data61 CSIRO)

Start year: 2022

Summary: Certified Deep Learning: Testing, Analysis and Verifying Deep Neural Networks. Deep learning techniques, especially deep neural networks (DNNs), have been widely used in various applications, such as image classification, natural language processing, software development, maintenance and bug detection. This PhD project aims to develop next-generation techniques to certify the robustness, correctness and safety of modern deep learning systems. Specifically, the project will leverage recent advances in program analysis and testing (e.g., recursive state machines, data flow analysis and differential testing) approaches to validate large-scale DNNs through traceable and precise adversarial samples. The research project will broadly cover but not limited: (1) Deep learning pipeline modeling via recursive state modeling (2) Robustness verification for DNNs through data flow analysis (3) Differential testing via precise and soundness-driven adversarial sample generation

FOR Codes: Application software packages, Formal methods for software, Automated software engineering

Acknowledgement of Country

UTS acknowledges the Gadigal people of the Eora Nation, the Boorooberongal people of the Dharug Nation, the Bidiagal people and the Gamaygal people, upon whose ancestral lands our university stands. We would also like to pay respect to the Elders both past and present, acknowledging them as the traditional custodians of knowledge for these lands.