Conference proceedings

ARTIFICIAL INTELLIGENCE-BASED DECISION MAKING, 
INFORMATION SECURITY A GAN-STYLE APPROACH IN 
CYBERATTACK DETECTION 
Azar Umudov* 1[0000-0002-2718-6330] 
, Tofiq Nazarov [0000-0002-3732-5832] 
and 
Sanan Cafarov [0000-0001-5693-437X] 
1, 2, 3 ASOIU, Baku, Azerbaijan 
[e-mail], [e-mail], 
[e-mail] 
Abstract.
Nowadays, the internet is being used extensively that it autogenously leads to encountering cyberattacks and 
malware. Variety techniques and systems have been applied to avoid and detect these anomalies in order to prevent their harmful 
effects on data security, software systems. Due to the success of deep neural networks in many domains, many models have been 
developed to detect cyberattacks. The main obstacle in real applications of these machine learning models is to collect labeled data 
samples. To overcome this problem, we developed a Semi-Supervised Generative Adversarial Network being trained with a small 
number of labeled data samples. We tested it over the NSL-KDD dataset andgained a successful result – 92.32% accuracy that was 
close to the performance of the studied models.