Sparse Mixture Once-for-all Adversarial Training for Efficient In-Situ Trade-Off Between Accuracy and Robustness of DNNs
Souvik Kundu (University of Southern California); Sairam Sundaresan (Intel AI Lab); Sharath Nittur Sridhar (Intel AI Lab); SHUNLIN LU (The Chinese University of Hong Kong); han Tang (University of Southern California); Peter A. Beerel (University of Southern California)
-
SPS
IEEE Members: $11.00
Non-members: $15.00
Existing deep neural networks (DNNs) that achieve state-of-the-art (SOTA) performance on both clean and adversarially-perturbed images rely on either activation or weight conditioned convolution operations. However, such conditional learning costs additional multiply-accumulate (MAC) or addition operations, increasing inference memory and compute costs. To that end, we present a sparse mixture once-for-all adversarial training (SMART), that allows a model to train once and then in-situ trade-off between accuracy and robustness, that too at a reduced compute and parameter overhead. In particular, SMART develops two expert paths, for clean and adversarial images, respectively, that are then conditionally trained via respective dedicated set of binary sparsity masks. Extensive evaluations on multiple image classification datasets across different models show SMART to have up to 2.72x fewer non-zero parameters costing proportional reduction in compute overhead, while yielding SOTA accuracy-robustness trade-off. Additionally, we present insightful observations in designing sparse masks to successfully condition on both clean and gradient-based perturbed images.