Open AccessQuest For Convergence Solution Using Hybrid Genetic Algorithm Trained Neural Network Model For Metamorphic Malware Detection
Author(s) -
Arnold Adimabua Ojugo,
Chris Obaro Obruche,
Andrew Okonji Eboka
Publication year - 2021
Publication title -
arrus journal of engineering and technology
Language(s) - English
Resource type - Journals
eISSN - 2807-3045
pISSN - 2776-7914
DOI - 10.35877/jetech613
Subject(s) - rationalization (economics) , algorithm , alarm , artificial neural network , genetic algorithm , computer science , malware , fitness function , recession , credit card , computer security , economics , artificial intelligence , machine learning , finance , engineering , microeconomics , payment , keynesian economics , aerospace engineering
An unstable economy is rife with fraud. Perpetrated on customers, it ranges from employees’ internal abuse to large fraud via high-value contracts cum control breaches that impose serious consequences to biz. Loyal employees may not perpetrate fraud if not for societal pressures and economic recession with its rationalization that they have bills to pay and children to feed. Thus, the need for financial institutions to embark on effective measures via schemes that will aids both fraud prevention and detection. Study proposes genetic algorithm trained neural net model to accurately classify credit card transactions. Compared, model used a rule-based system to provide it with start-up solution and it has a fraud catching rate of 91% with a consequent, false alarm rate of 9%. Its convergence time is found to depend on how close the initial solution space is to the fitness function, and for recombination and mutation rates applied.