CANCELLED: RoHNAS–When Robustness Meets HW-Aware NAS: A Neural Architecture Search Framework with Conjoint Optimization for Adversarial Robustness and Hardware Efficiency of Convolutional and Capsule Networks
TimeWednesday, December 8th6:00pm - 7:00pm PST
LocationLevel 2 - Lobby
Event Type
Networking Reception
Work-in-Progress Poster
Virtual Programs
Presented In-Person
DescriptionNeural Architecture Search (NAS) algorithms aim at finding high accurate Deep Neural Network (DNN) architectures for a given application. However, DNNs are computationally-complex as well as vulnerable to adversarial attacks. Towards this, we propose RoHNAS, a novel NAS framework that jointly optimizes for the adversarial-robustness and hardware-efficiency of DNNs executed on specialized hardware accelerators. RoHNAS additionally accounts for complex types of DNNs such as Capsule Networks. RoHNAS analyzes and selects values of adversarial perturbation for each dataset to employ in the NAS. Our evaluations provide a set of Pareto-optimal solutions, leveraging the tradeoff between the above-discussed design objectives.