Authors :
Hafsat Bida Abdullahi
Volume/Issue :
Volume 9 - 2024, Issue 2 - February
Google Scholar :
http://tinyurl.com/ywnbabrp
Scribd :
http://tinyurl.com/yc6kbh28
DOI :
https://doi.org/10.5281/zenodo.10670055
Abstract :
Cyberattacks are evolving, and conventional
signature-based detection mechanisms will not succeed at
detecting such attacks. Sophisticated detection systems
that utilize modern data analytics, such as machine
learning and artificial intelligence, can identify hidden
patterns or behavioral relationships in the large array of
cyber-related residuals. This study suggests cyber threat
detection research into a comprehensive artificial
intelligence framework. The features should have behavior
modeling, intelligent correlation, and dynamic detection
models. All these difficulties are the challenges to human
research efforts as related to new endeavors with multi-
source data sets. They also include three different, most
optimized algorithms with chances of being free from such
production variants that are biased multi-mode sources.
With the constant informing of realistic threats, machine
learning models have to produce sturdy representations
that can transfer knowledge to identify innovative attacks.
Transparency and auditability of a model encourage faith
in automated decisions. Continual training against
adversarial samples and concept drift makes them
resilient. End-to-end, multi-layered cyber defense benefits
from a variety of sources, including integrated analytics
leveraging the full spectrum visibility through
orchestration across the network, user, and malware data.
The alternative learning paradigms of self-supervision and
reinforcement learning provide hope to topics such as
high-valued threat intelligence. Finally, human-machine
integration, which takes advantage of strengths based on
complementary aptitudes, shall chart the next course.
Analyst cognition-enhancing algorithms decrease
operational workloads. The scope of this study is to
promote cyber protection with A.I. evolving beyond
traditional limitations.
Cyberattacks are evolving, and conventional
signature-based detection mechanisms will not succeed at
detecting such attacks. Sophisticated detection systems
that utilize modern data analytics, such as machine
learning and artificial intelligence, can identify hidden
patterns or behavioral relationships in the large array of
cyber-related residuals. This study suggests cyber threat
detection research into a comprehensive artificial
intelligence framework. The features should have behavior
modeling, intelligent correlation, and dynamic detection
models. All these difficulties are the challenges to human
research efforts as related to new endeavors with multi-
source data sets. They also include three different, most
optimized algorithms with chances of being free from such
production variants that are biased multi-mode sources.
With the constant informing of realistic threats, machine
learning models have to produce sturdy representations
that can transfer knowledge to identify innovative attacks.
Transparency and auditability of a model encourage faith
in automated decisions. Continual training against
adversarial samples and concept drift makes them
resilient. End-to-end, multi-layered cyber defense benefits
from a variety of sources, including integrated analytics
leveraging the full spectrum visibility through
orchestration across the network, user, and malware data.
The alternative learning paradigms of self-supervision and
reinforcement learning provide hope to topics such as
high-valued threat intelligence. Finally, human-machine
integration, which takes advantage of strengths based on
complementary aptitudes, shall chart the next course.
Analyst cognition-enhancing algorithms decrease
operational workloads. The scope of this study is to
promote cyber protection with A.I. evolving beyond
traditional limitations.
