No abstract

Recent studies identify that Deep learning Neural Networks (DNNs) are vulnerable to subtle perturbations, which are not perceptible to human visual system but can fool the DNN models and lead to wrong outputs. A class of adversarial attack network algorithms has been proposed to generate robust physical perturbations under different circumstance. These algorithms are the first efforts to move forward secure deep learning by providing an avenue to train future defense networks, however, the intrinsic complexity of them prevents their broader usage. In this paper, we propose the first hardware accelerator for adversarial attacks based on memristor crossbar arrays. Our design significantly improves the throughput of a visual adversarial perturbation system, which can further improve the robustness and security of future deep learning systems. Based on the algorithm uniqueness, we propose four implementations for the adversarial attack accelerator (A 3) to improve the throughput, energy efficiency, and computational efficiency.

No abstract

Vigorously promoting the development of photovoltaic (PV) resources is a positive measure taken by humanity in response to the changes in global climate and environment. At the same time, combining photovoltaic power generation systems with traditional power generation systems, using the advantages of different power generation systems to achieve real-time scheduling optimization has become an urgent problem to be solved in engineering applications. This paper attempts to study the climate and environmental benefits of the development of photovoltaic resource in Africa by taking Angola as an example based on actual project data. According to the characteristics, load requirements, seasonal characteristics and actual engineering background of Tombwa in Angola, a baseline Scenario and four comparative Scenarios were established, and the operating costs of the five Scenarios in local rainy season and dry season were obtained respectively. The cost of electricity for the five Scenarios calculated subsequently. Through real-time scheduling and optimization of the software, the emission characteristics of CO2, NOx and CO under five Scenarios are obtained, and the climate benefits and environmental benefits of the five scenarios are further analyzed and compared. The results show that the development of photovoltaic resources in Angola has good climate and environmental benefits. In addition, the combine application of diesel, PV and battery power system will be the most effective of the five Scenarios to reduce the CO2 emissions with the lowest levelized cost of electricity (LCOE) of 0.38 yuan/kwh, as a cost-effective solution in remote areas of Angola, Africa.