Author: Community

A Cooperative Co-Evolutionary Approach to Large-Scale Multisource Water Distribution Network Optimization

Potable water distribution networks (WDNs) are important infrastructures of modern cities. A good design of the network can not only reduce the construction expenditure but also provide reliable service. Nowadays, the scale of the WDN of a city grows d…

Potable water distribution networks (WDNs) are important infrastructures of modern cities. A good design of the network can not only reduce the construction expenditure but also provide reliable service. Nowadays, the scale of the WDN of a city grows dramatically along with the city expansion, which brings heavy pressure to its optimal design. In order to solve the large-scale WDN optimization problem, a cooperative co-evolutionary algorithm is proposed in this paper. First, an iterative trace-based decomposition method is specially designed by utilizing the information of water tracing to divide a large-scale network into small subnetworks. Since little domain knowledge is required, the decomposition method has great adaptability to multiform networks. Meanwhile, during optimization, the proposed algorithm can gradually refine the decomposition to make it more accurate. Second, a new fitness function is devised to handle the pressure constraint of the problem. The function transforms the constraint into a part of the objective to punish the infeasible solutions. Finally, a new suite of benchmark networks are created with both balanced and imbalanced cases. Experimental results on a widely used real network and the benchmark networks show that the proposed algorithm is promising.

One Pixel Attack for Fooling Deep Neural Networks

Recent research has revealed that the output of deep neural networks (DNNs) can be easily altered by adding relatively small perturbations to the input vector. In this paper, we analyze an attack in an extremely limited scenario where only one pixel ca…

Recent research has revealed that the output of deep neural networks (DNNs) can be easily altered by adding relatively small perturbations to the input vector. In this paper, we analyze an attack in an extremely limited scenario where only one pixel can be modified. For that we propose a novel method for generating one-pixel adversarial perturbations based on differential evolution (DE). It requires less adversarial information (a black-box attack) and can fool more types of networks due to the inherent features of DE. The results show that 67.97% of the natural images in Kaggle CIFAR-10 test dataset and 16.04% of the ImageNet (ILSVRC 2012) test images can be perturbed to at least one target class by modifying just one pixel with 74.03% and 22.91% confidence on average. We also show the same vulnerability on the original CIFAR-10 dataset. Thus, the proposed attack explores a different take on adversarial machine learning in an extreme limited scenario, showing that current DNNs are also vulnerable to such low dimension attacks. Besides, we also illustrate an important application of DE (or broadly speaking, evolutionary computation) in the domain of adversarial machine learning: creating tools that can effectively generate low-cost adversarial attacks against neural networks for evaluating robustness.