1758
ResearchBib IF - 11.01, ISSN: 3030-3753, Volume 2 Issue 5
SUN’IY INTELLEKT YORDAMIDA KIBERXAVFSIZLIKNI MUSTAHKAMLASH:
ZAMONAVIY YONDASHUVLAR VA ALGORITMLAR
Eshmurodov Mas’udjon Xikmatillayevich
Samarqand davlat arxitektura-qurilish universiteti, 140147, Samarqand, Uzbekistan.
ORCID ID:
https://orcid.org/0009-0005-0667-8116
masudeshmurodov@samdaqu.edu.uz
, +998933501484.
Shaimov Komiljon Mirzakabulovich
ORCID ID:
https://orcid.org/0009-0005-8279-4530
shaimovkomiljon@gmail.com
, +998937228187.
Gaybulov Qodirjon Murtozoyevich
ORCID ID:
https://orcid.org/0000-0001-9575-0338
q.gaybulov@samdaqu.edu.uz,
+998885059905.
Elmurodov Bahodir Ergashevich
ORCID ID:
https://orcid.org/0009-0003-6390-7961
+998912976135.
https://doi.org/10.5281/zenodo.15558910
Annotatsiya. Ushbu maqolada sun’iy intellekt (SI) texnologiyalarining kiberxavfsizlik
sohasiga tatbiq etilishi, ayniqsa zamonaviy yondashuvlar va mashinaviy o‘rganish algoritmlari
asosida tahdidlarni aniqlash, bashorat qilish va ularga javob berish bo‘yicha imkoniyatlari
yoritiladi. Shuningdek, amaliy misollar va natijalar asosida SI vositalarining samaradorligi
tahlil qilinadi.
Kalit so‘zlar: Sun’iy intellekt, kiberxavfsizlik, mashinaviy o‘rganish, tahdid aniqlash,
algoritmlar, himoya tizimlari.
Abstract. This article explores the application of artificial intelligence (AI) technologies
in the field of cybersecurity, with a focus on modern approaches and machine learning
algorithms used to detect, predict, and respond to threats. It also analyzes the effectiveness of AI
tools based on practical examples and results.
Keywords: Artificial intelligence, cybersecurity, machine learning, threat detection,
algorithms, defense systems.
1. Kirish (Introduction)
Axborot texnologiyalarining keng rivojlanishi bilan bir qatorda kiberxavf-xatarlar ham
tezlik bilan ortib bormoqda. Har yili yuzlab kompaniyalar va foydalanuvchilar zarar ko‘rmoqda.
An’anaviy kiberxavfsizlik tizimlari bu tahdidlarning murakkabligi va tez
o‘zgaruvchanligiga qarshi yetarli darajada samarali emas. Shu bois, oxirgi yillarda
sun’iy
intellekt (SI)
va
mashinaviy o‘rganish
algoritmlaridan foydalanish kiberxavfsizlikni
mustahkamlashda muhim vositaga aylanmoqda. SI yordamida tizimlar nafaqat mavjud
tahdidlarni aniqlaydi, balki yangi, ilgari noma’lum bo‘lgan hujumlarni ham bashorat qila oladi.
2. Usullar (Methods)
Ushbu maqolani tayyorlashda quyidagi SI usullari va algoritmlarining kiberxavfsizlikda
qo‘llanishi o‘rganildi:
•
Mashinaviy o‘rganish (ML)
: tahdidlarni aniqlash va tahlil qilish uchun tasniflash
(classification) va klasterlash (clustering) metodlari.
•
Chuqur o‘rganish (Deep Learning)
: neyron tarmoqlar orqali kiberhujumlarni real vaqt
rejimida aniqlash.
1759
ResearchBib IF - 11.01, ISSN: 3030-3753, Volume 2 Issue 5
•
Anomaliyani aniqlash algoritmlari
: foydalanuvchi xatti-harakatlaridagi o‘zgarishlar
orqali kiberxurujlarni oldindan ko‘rish.
•
NLP (Natural Language Processing)
: phishing xabarlarini va zararli e-mail’larni
aniqlash.
Tadqiqot davomida ochiq ma’lumotlar to‘plami (NSL-KDD dataset, CIC-IDS2017)
asosida modellarning aniqlik darajasi (accuracy), sezgirlik (sensitivity) va xatolik darajasi (false
positive rate) tahlil qilindi.
3. Natijalar (Results – Expanded)
Tadqiqot davomida bir nechta sun’iy intellekt (SI) texnologiyalari va algoritmlarining
samaradorligi turli kiberxavfsizlik vazifalari bo‘yicha tahlil qilindi va quyidagi natijalar olindi.
a)
Tasniflovchi modellarning natijalari
Tasniflovchi algoritmlar ichida Random Forest eng yuqori natijani ko‘rsatdi. NSL-KDD
ma’lumotlar to‘plami asosida o‘tkazilgan tajribalarda ushbu model:
•
96% aniqlik (accuracy)
•
94% sezgirlik (recall)
•
3.8% noto‘g‘ri ijobiy natija (false positive rate)
ko‘rsatdi.
Bu model ayniqsa DDoS hujumlari
,
port scanning
va
backdoor urinishlarini aniqlashda
barqaror natijalarga erishdi.
b)
Chuqur o‘rganish (Deep Learning – LSTM) natijalari
Long Short-Term Memory (LSTM) asosidagi chuqur o‘rganish modelidan anomaliyani
aniqlash vazifasida foydalanildi. Bu model real vaqt rejimida tarmoq oqimi (network traffic)
asosida tahdidlarni aniqlashga mo‘ljallangan.
•
93% ishonchlilik (confidence)
•
91% aniqlik
•
2.5% noto‘g‘ri rad etish (false negative) holatlari kuzatildi.
Model
xususan
vaqt
ketma-ketligi
asosida
foydalanuvchi
xatti-harakatlaridagi
o‘zgarishlarni muvaffaqiyatli aniqladi. Bu ayniqsa insider threats va zero-day attacks uchun
foydalidir.
c)
NLP asosidagi phishing aniqlash tizimi
Natural Language Processing (NLP) texnologiyasi asosida tuzilgan model zararli
xabarlarni (e-mail, matn) tahlil qildi. Bu yerda text classification va semantic analysis
yondashuvlari qo‘llanildi.
•
92% aniqlik
•
Yuqori darajadagi kontekstual aniqlash
(grammar-based detection)
•
URL, matn va yozuvdagi psixologik manipulyatsiya uslublarini aniqlashga qodir bo‘ldi.
Shu bilan birga, soxta havolalar va ijtimoiy muhandislik (social engineering) asosidagi
hujumlar muvaffaqiyatli aniqladi.
d)
O‘z-o‘zini o‘rganuvchi tizimlar (Self-learning systems)
Adaptiv (moslashuvchan) algoritmlar — xususan reinforcement learning
va
semi-
supervised learning asosidagi modellar — kiberxavf-xatarlarning o‘zgaruvchan tabiatiga yaxshi
moslashuvchanlik ko‘rsatdi.
•
Dinamik hujum modellarini aniqlashda ustunlikka ega bo‘ldi.
•
Agar ma’lumotlar yangilanmasa ham, model ilgari o‘rgangan tajribalar asosida qaror
chiqarishda davom etdi.
1760
ResearchBib IF - 11.01, ISSN: 3030-3753, Volume 2 Issue 5
•
Aytish mumkinki, bunday tizimlar kiberxavfsizlikni avtomatlashtirish va real vaqtli qaror
qabul qilishda muhim ahamiyatga ega.
4. Munozara (Discussion)
Natijalar shuni ko‘rsatadiki, sun’iy intellekt asosidagi yondashuvlar kiberxavfsizlik
sohasida yuqori samaradorlikka ega. Ayniqsa, real vaqt rejimida ishlovchi chuqur o‘rganish
tizimlari foydalanuvchi xatti-harakatlarining noan’anaviy o‘zgarishlarini aniqlashda foydalidir.
Shu bilan birga, SI texnologiyalari hali ham quyidagi cheklovlarga ega:
•
Soxta musbat natijalar (false positives) soni yuqori bo‘lishi mumkin.
•
Ma’lumotlar maxfiyligi va etik muammolar yuzaga chiqadi.
•
Modellarni tushunish qiyinligi (black-box muammosi) ba’zi holatlarda muhim qarorlarni
asoslashni qiyinlashtiradi.
Kelajakda SI algoritmlari yanada optimallashtirilgan, tushunarli va ishonchli bo‘lishi
lozim.
Xulosa
Sun’iy intellekt yordamida kiberxavfsizlikni mustahkamlash zamonaviy axborot
muhofazasi tizimlarining muhim yo‘nalishiga aylangan. SI asosidagi modellar an’anaviy
yondashuvlarga qaraganda yuqoriroq aniqlik, moslashuvchanlik va tezkorlikni ta’minlaydi. Shu
bilan birga, bu texnologiyalarni ehtiyotkorlik bilan, axloqiy va huquqiy me’yorlarga amal qilgan
holda joriy etish muhim.
REFERENCES
1.
Buczak, A. L., & Guven, E. (2016). A survey of data mining and machine learning
methods for cyber security intrusion detection
.
IEEE Communications Surveys &
Tutorials
, 18(2), 1153–1176. DOI: 10.1109/COMST.2015.2494502
2.
Chio, C., & Freeman, D. (2018).
Machine Learning and Security: Protecting Systems
with Data and Algorithms.
O’Reilly Media.
3.
Sarker, I. H. (2022).
Machine learning techniques for cybersecurity: A comprehensive
review
and
roadmap.
IEEE
Access
,
9,
17265–17299.
DOI: 10.1109/ACCESS.2021.3135954
4.
Sharafaldin, I., Lashkari, A. H., & Ghorbani, A. A. (2018).
Toward generating a new
intrusion detection dataset and intrusion traffic characterization.
ICISSP
, 108–116.
(CIC-IDS2017 dataset asoschisi)
5.
Javaid, A., Niyaz, Q., Sun, W., & Alam, M. (2016).
A deep learning approach for
network intrusion detection system.
Proceedings of the 9th EAI International Conference
on Bio-inspired Information and Communications Technologies (BICT 2015)
.
DOI: 10.4108/eai.3-12-2015.2262516
6.
Kaspersky Lab (2023).
Artificial Intelligence in Cybersecurity: Opportunities and Risks.
7.
Sommer, R., & Paxson, V. (2010).
Outside the closed world: On using machine learning
for network intrusion detection.
2010 IEEE Symposium on Security and Privacy
, 305–316.
DOI: 10.1109/SP.2010.25
8.
Goodfellow, I., Bengio, Y., & Courville, A. (2016).
Deep Learning.
MIT Press.
(Chuqur o‘rganish nazariyasi uchun asosiy manba)
9.
Zhang, Y., & Paxson, V. (2020).
Detecting and analyzing automated activity on large-
scale network infrastructures.
ACM Transactions on Privacy and Security
, 23(1), 1–30.
1761
ResearchBib IF - 11.01, ISSN: 3030-3753, Volume 2 Issue 5
10.
NSL-KDD
Dataset
–
University
of
New
Brunswick.
https://www.unb.ca/cic/datasets/nsl.html
11.
Шаимов К.М., Эшмуродов М.Х., Хужаев И.К. Дифференциально-разностный метод
для двумерных линейных задач теплопередачи // Научный вестник. СамГУ – 2020, –
№1(121). – C.78-87(01.00.00.; № 2).
12.
M Kh Eshmurodov, K.M. Shaimov, I Khujaev and J Khujaev Method of lines for solving
linear equations of mathematical physics with the third and first types boundary conditions.
Journal of Physics: Conference Series 2131 (2021) 032041, doi:10.1088/1742-
6596/2131/3/032041
13.
K. M. Shaimov, M. Kh. Eshmurodov, I. Khujaev and Zh. I. Khujaev The Method of Lines
for Solving Equations of Mathematical Physics with Boundary Conditions of the First and
Third Types // The method of lines for solving equations of mathematical physics with
boundary conditions of the first and third types, Cite as: AIP Conference Proceedings
2612, 030028 (2023);
https://doi.org/10.1063/5.0124614
, Published Online: 15 March
2023
14.
K. M. Shaimov, M. Kh. Eshmurodov, M.T. Shodmonqulov, Q.M. Gaybulov. Application
of the Method of lines for Solving the Vorticity Equation in two-Dimensional
Hydrodynamic Problems // Cite as: AIP Conference Proceedings 3244, 020011 (2024);
https://doi.org/10.1063/5.0242469
, Published Online: 27 November 2024
