Adversarial-Aware Deep Learning System Based on a Secondary Classical Machine Learning Verification Approach.

Deep learning models have been used in creating various effective image classification applications. However, they are vulnerable to adversarial attacks that seek to misguide the models into predicting incorrect classes. Our study of major adversarial attack models shows that they all specifically target and exploit the neural networking structures in their designs. This understanding led us to develop a hypothesis that most classical machine learning models, such as random forest (RF), are immune to adversarial attack models because they do not rely on neural network design at all. Our experimental study of classical machine learning models against popular adversarial attacks supports this hypothesis. Based on this hypothesis, we propose a new adversarial-aware deep learning system by using a classical machine learning model as the secondary verification system to complement the primary deep learning model in image classification. Although the secondary classical machine learning model has less accurate output, it is only used for verification purposes, which does not impact the output accuracy of the primary deep learning model, and, at the same time, can effectively detect an adversarial attack when a clear mismatch occurs. Our experiments based on the CIFAR-100 dataset show that our proposed approach outperforms current state-of-the-art adversarial defense systems.

Source to Tap Risk Assessment for Intermittent Water Supply Systems in Arid Regions: An Integrated FTA-Fuzzy FMEA Methodology.

Water utilities in arid regions deal with multifaceted issues of natural groundwater contamination, high treatment costs, and low water rates. These utilities rely on intermittent supplies resulting in numerous water quality failures at source, treatment, distribution, and in-house plumbing systems. The present research presents an inclusive risk assessment methodology for managing water quality from source to tap. Three-year monitoring data for turbidity, TDS, pH, iron, ammonia, nitrates, residual chlorine, Coliform group, E. coli, and Fecal Streptococci identified the root causes of failures. The cause-effect relationships in the form of a fault tree were solved using multiple failure modes and effect analysis (FMEA) to handle both the Boolean operations. The fuzzy sets addressed the uncertainties associated with data limitations in calculating exceedance probabilities (Pe) and vagueness in expert opinion for subjective evaluation of severity and detectability. The methodology was applied on a smaller system serving 18,000 consumers in Qassim, Saudi Arabia. Potable supplied water underwent reoccurrence of TDS (Pe = 20%), turbidity (Pe = 10%), and Fe (Pe = 2%) failures in distribution that further increased up to 44%, 33%, and 11% at the consumer end. The Pe for residual chlorine failure soared up to 89%. Economic controls reduced the cumulative risk to 50%, while the shift to continuous supply can limit the remaining failures under the acceptable risk. The framework will help utilities manage water quality in intermittent systems from source to tap in Saudi Arabia, the Gulf, and elsewhere.