Sehla Loussaief, Afef Abdelkrim
Adv. Sci. Technol. Eng. Syst. J. 3(1), 01-10 (2018);
Hereby in this paper, we are going to refer image classification. The main issue in image classification is features extraction and image vector representation. We expose the Bag of Features method used to find image representation. Class prediction accuracy of varying classifiers algorithms is measured on Caltech 101 images. For feature extraction functions we evaluate the use of the classical Speed Up Robust Features technique against global color feature extraction. The purpose of our work is to guess the best machine learning framework techniques to recognize the stop sign images. The trained model will be integrated into a robotic system in a future work.
Hassan Aziza, Pierre Canet, Jeremy Postel-Pellerin
Adv. Sci. Technol. Eng. Syst. J. 3(1), 11-17 (2018);
In this paper, the performance and reliability of oxide-based Resistive RAM (ReRAM) memory is investigated in a 28nm FDSOI technology versus interconnects resistivity combined with device variability. Indeed, common problems with ReRAM are related to high variability in operating conditions and low yield. At a cell level ReRAMs suffer from variability. At an array level, ReRAMs suffer from different voltage drops seen across the cells due to line resistances. Although research has taken steps to resolve these issues, variability combined with resistive paths remain an important characteristic for ReRAMs. In this context, a deeper understanding of the impact of these characteristics on ReRAM performances is needed to propose variability tolerant designs to ensure the robustness of the technology. The presented study addresses the memory cell, the memory word up to the memory matrix.
El Mostapha Chakir, Mohamed Moughit, Youness Idrissi Khamlichi
Adv. Sci. Technol. Eng. Syst. J. 3(1), 18-24 (2018);
This paper is an extension of work originally presented in WITS-2017 CONF. We extend our previous works by improving the Risk calculation formula, and risk assessment of an alert cluster instead of every single alert. Also, we presented the initial results of the implementation of our model based on risk assessment and alerts prioritization. The idea focuses on a new approach to estimate the risk of each alert and a cluster of alerts. This approach uses indicators such as priority, reliability and asset value as decision factors to calculate alert’s risk. The objective is to determine the impact of alerts generated by Intrusion detection system (IDS) on the security status of an information system, and also improve the detection of intrusions using snort IDS by classifying the most critical alerts by their levels of risk. Thus, only alerts that present a real threat will be displayed to the security administrator. The implementation of this approach will reduce the number of false alerts and improve the performance of the IDS.
Himanshu Upadhyay, Hardik Gohel, Alexander Pons, Leo Lagos
Adv. Sci. Technol. Eng. Syst. J. 3(1), 25-29 (2018);
In today’s information based world, it is increasingly important to safeguard the data owned by any organization, be it intellectual property or personal information. With ever increasing sophistication of malware, it is imperative to come up with an automated and advanced methods of attack vector recognition and isolation. Existing methods are not dynamic enough to adapt to the behavioral complexity of new malware. Widely used operating systems, especially Linux, have a popular perception of being more secure than other operating systems (e.g. Windows), but this is not necessarily true. The open source nature of the Linux operating system is a double edge sword; malicious actors having full access to the kernel code does not reassure the IT world of Linux’s vulnerabilities. Recent widely reported hacking attacks on reputable organizations have mostly been on Linux servers. Most new malwares are able to neutralize existing defenses on the Linux operating system. A radical solution for malware detection is needed – one which cannot be detected and damaged by malicious code. In this paper, we propose a novel framework design that uses virtualization to isolate and monitor Linux environments. The framework uses the well-known Xen hypervisor to host server environments and uses a Virtual Memory Introspection framework to capture process behavior. The behavioral data is analyzed using sophisticated machine learning algorithms to flag potential cyber threats. The framework can be enhanced to have self-healing properties: any compromised hosts are immediately replaced by their uncompromised versions, limiting the exposure to the wider enterprise network.