Reporting by Alli Weaver.
The grant, given by the Division of Computer and Network Systems within NSF, was awarded for Gong to investigate mitigating evasion attacks on neural networks.
“Deep neural networks have transformed several artificial intelligence research areas, including computer vision, speech recognition and natural language processing,” Gong states. “However, recent studies demonstrated that DNNs are vulnerable to adversarial manipulations at testing time.”
Some initial foundational work for this grant was published in Gong’s 2017 article, “Mitigating Evasion Attacks to Deep Neural Networks via Region-based Classification.”
According to Gong, an attacker can add a noise to the testing sample that the DNNs use to predict, causing an evasion attack, which is one of the biggest challenges in safety and security of DNNs, such as in self-driving cars.
Gong, along with Xiaoyu Cao, ECpE graduate student, developed new DNNs, which would deter against the evasion attacks.
The research was published in ACSAC 2017 Proceedings of the 33rd Annual Computer Security Applications Conference. Though 231 submissions were made for this conference, only 44 were accepted.
Gong has been an assistant professor at Iowa State since fall 2015, after receiving his Ph.D. from the University of California at Berkeley and his bachelor’s in computer science from the University of Science and Technology of China in 2010.
Gong received an NSF Career Award in 2018.