Adversarial Training in Stochastic Bitwise Neural Networks: Robustness Analysis Against FGSM Attacks on CIFAR-10

Recently published methods enable training of bitwise neural networks which allow reduced representation of down to a single bit per weight. We present a method that exploits ensemble decisions based on multiple stochastically sampled network models to increase performance figures of bitwise neural networks in terms of classification accuracy at inference. Our experiments with the CIFAR-10 and GTSRB datasets show that the performance of such network ensembles surpasses the performance of the high-precision base model. With this technique we achieve 5.81\% best classification error on CIFAR-10 t

Research goal: How does adversarial training with stochastic sampling in bitwise neural networks impact robustness against FGSM attacks on CIFAR-10 compared to deterministic weight-based BNNs, measured by classification accuracy under varying perturbation magnitudes?

Autonomous synthesis report generated by SOVEREIGN Research Kernel. Tribunal consensus score: 8.7/10.