Sponge attacks aim to increase the energy consumption and computation time of neural networks deployed on hardware accelerators. Existing sponge attacks can be performed during inference via sponge examples or during training via Sponge Poisoning. Sponge examples leverage perturbations added to the model's input to increase energy and latency, while Sponge Poisoning alters the objective function of a model to induce inference-time energy effects. In this work, we propose a novel sponge attack called SkipSponge. SkipSponge is the first sponge attack that is performed directly on the parameters of a pre-trained model using only a few data samples. Our experiments show that SkipSponge can successfully increase the energy consumption of image classification models with fewer samples required than Sponge Poisoning. We show that poisoning defenses are ineffective if not adjusted specifically for the defense against SkipSponge (i.e., they decrease target layer bias values). Our work shows that SkipSponge is more effective on the GANs and the autoencoders than the state-of-the-art. Additionally, SkipSponge is stealthier than the previous Sponge Poisoning attack as it does not require significant changes in the victim model's weights. Our experiments indicate that the SkipSponge attack can be performed even when an attacker has access to only 1% of the entire dataset and reaches up to 13% energy increase.

翻译：暂无翻译