Specialized Genetic Algorithm Based Simulation Tool Designed For Malware Evolution Forecasting

Abstract

From the security point of view malware evolution forecasting is very important, since it provides an opportunity to predict malware epidemic outbreaks, develop effective countermeasure techniques and evaluate information security level. Genetic algorithm approach for mobile malware evolution forecasting already proved its effectiveness. There exists a number of simulation tools based on the Genetic algorithms, that could be used for malware forecasting, but their main disadvantages from the user's point of view is that they are too complicated and can not fully represent the security entity parameter set. In this article we describe the specialized evolution forecasting simulation tool developed for security entities, such as different types of malware, which is capable of providing intuitive graphical interface for users and ensure high calculation performance. Tool applicability for the evolution forecasting tasks is proved by providing mobile malware evolution forecasting results and comparing them with the results we obtained in 2010 by means of MATLAB. *

Survey of machine learning techniques for malware analysis

“…Howard et al This problem can be seen as yet another objective in the malware analysis taxonomy. It has not been investigated much yet, only a couple of works seem to address that topic [103,102]. Given its great potential to proactively identify novel malware, and considering the opportunity to exploit existing malware families datasets, we claim the worthiness to boost the research on malware evolution prediction through machine learning techniques.…”

Qualities, Challenges and Future of Genetic Algorithms

“…A plausible future of GAs lies in the modelling of continuously-adaptive systems. In the same way GAs have been used by Juzonis et al, 2012 to simulate malware evolution to forecast malware epidemic outbreaks, and contribute to the improvement of responses and threat evaluations, having models in economics, finance, but also other fields, that simulate the quasi open-ended evolution of human behavior, can shed new light on the behavior of these systems. Understanding the behavior effects of sudden changes in stock markets, improving our stress testing models to include the deep transitions in agents' behaviors, are means to achieve more robust and more efficient economies.…”

“…The encoding representation indirectly constrains the search space, with the issue of partial cover. Only a robust, careful representation design allows to cover its fully diversity (Juzonis et al, 2012). A founding analysis on encoding in Genetic algorithms has been done by Ronald, 1997.…”

Artificial Intelligence Based Malware Analysis

“…This type of prediction may not be of practical use to a malware analyst, as they are not able to quantify the threat level of a malware family or trend. A number of tools and systems [29,30,31] have been proposed for detecting novelty or anomaly among individual malware binaries. These do not include the notion of a malware family, do not consider changes in the families over time, nor do they make long term predictions about the behavior of a family over time.…”