Machine learningSyllabus

Subject code

İFS-2219yn

Credits

6

Student workload

120

Semester

Autumn semester

Academic year

4

Subject type

Elective

Form of education

Full-time

Language of instruction

Azerbaijani

In-class hours

a) 30 saat - mühazirə b) 15 saat - seminar c) 15 saat - laboratoriya

Out-of-class hours

a) 60 saat - TSİ b) 60 saat - MRTSİ

The subject provides students with the formation of knowledge and practical skills on the theoretical foundations of machine learning, basic algorithms and their application in the field of information security. within the subject, the basic concepts of machine learning, data preparation, pre-processing and feature engineering methods are studied. the working principles and application characteristics of classical machine learning algorithms such as linear regression, decision trees, support vector machines (svm), k-nearest neighbor (knn), and clustering are reviewed. the application of these methods to information security issues such as prediction, anomaly and attack detection, identification of malicious activity patterns, and clustering of network attacks is explored. also studying malware and network traffic analysis based on artificial neural networks, deep learning and convolutional neural networks, detection of spam, phishing and social engineering attacks using natural language processing (nlp) methods. model evaluation and optimization methods, security risks and opportunities created by generative artificial intelligence, the use of artificial intelligence agents in security systems, as well as ethical and social aspects of artificial intelligence technologies are reviewed. the subject is aimed at developing students' skills in applying machine learning algorithms to specific information security problems, evaluating the results of models and choosing artificial intelligence approaches in accordance with security requirements.

• lecture
• laboratory work
• presentations
• seminar
• problem-based learning
• interactive discussion

1. 1

   CLO 1: Maşın öyrənməsinin əsas konseptlərini, nəzarətli və nəzarətsiz öyrənmə alqoritmlərinin riyazi-məntiqi əsaslarını və müasir si̇ yanaşmalarını izah edir və müqayisə edir.

2. 2

   CLO 2: Verilənlərin təmizlənməsi, çatışmayan dəyərlərin bərpası və əlamət mühəndisliyi (feature engineering) üsullarını real məlumat dəstlərinə tətbiq edir.

3. 3

   CLO 3: Təbii dilin emalı (nlp), kompüter görməsi (cnn) və generativ si̇ sahələrində spesifik məsələlərin həlli üçün maşın öyrənməsi və dərin öyrənmə alqoritmlərini analiz edir və müvafiq modeli müəyyənləşdirir.

4. 4

   CLO 4: Qurulmuş maşın öyrənməsi modellərinin performansını müvafiq metrikalarla qiymətləndirir, həddindən artıq uyğunlaşma (overfitting) riskini idarə edir və hiperparametrləri optimallaşdırır.

5. 5

   CLO 5: Maşın öyrənməsi sistemlərinin tətbiqi zamanı ortaya çıkan etik prinsipləri, məfhumluq və sosial təsirləri qiymətləndirir və təhlükəsiz si̇ həlli təqdim edir.

1. 1.
   Introduction to machine learningLecture

   Within the framework of the topic, the main concepts, goals and application areas of machine learning in information security are studied. the main methods used in malware analysis and the role of machine learning methods in this process are reviewed. major types of machine learning, such as supervised, unsupervised, and reinforcement learning, are explored, their properties, and their application in malware detection and classification.

2. 2.
   Data preparation and feature engineeringLecture

   Within the framework of the topic, the role and main stages of data preparation in the process of machine learning are studied. key features of security data such as logs, network traffic, and event logs are reviewed. methods for cleaning data, handling null values, and removing duplicate records are explained. the basic principles of feature selection and feature selection methods for machine learning models are also explored.

3. 3.
   Linear regressionLecture

   Within the subject, the concept of linear regression, its main purpose and role in machine learning are studied. principles of modeling linear relationships between dependent and independent variables, building regression models, and forecasting are reviewed. the main stages of regression analysis and interpretation of model results are also explored.

4. 4.
   Decision treesLecture

   Within the topic, the concept of decision trees, their main working principles and their role in machine learning are studied. the structure of a decision tree consisting of root node, internal nodes, branches and leaf nodes is considered. the principles of data segmentation based on characteristics and selection of appropriate segmentation criteria are explained. the application of decision trees in the detection and classification of attacks, anomalies and malicious activity patterns in information security is also investigated.

5. 5.
   Support vector machines (svm)Lecture

   Within the framework of the topic, the concept of support vector machines (svm), its main purpose and application in classification problems are studied. determining the optimal boundary separating classes and the role of support vectors in model building are considered. a brief introduction to basic kernel functions such as linear, polynomial, and radial basis function (rbf) is given. also, one-vs-rest and one-vs-one approaches for solving multi-class classification problems are investigated.

6. 6.
   Knn və nümunə əsaslı metodlarLecture

   Mövzu çərçivəsində k-ən yaxın qonşu (knn) alqoritmi və nümunə əsaslı öyrənmənin (instance-based learning) əsas prinsipləri öyrənilir. KNN alqoritminin yeni verilənləri mövcud nümunələrlə oxşarlıq və məsafə əsasında qiymətləndirməsi, ən yaxın qonşuların müəyyən edilməsi və (k) parametrinin seçilməsi prinsipləri nəzərdən keçirilir. həmçinin KNN alqoritminin təsnifat və reqressiya məsələlərində iş prinsipi və tətbiq xüsusiyyətləri araşdırılır.

   • • род 1: объяснять основные характеристики периодов истории азербайджана.
   • • род 4: демонстрировать базовый уровень знаний по изучаемым темам и формулировать собственную аргументированную позицию по рассматриваемым вопросам.
   • • род 2: анализировать и разъяснять отдельные дискуссионные вопросы истории азербайджана.
   • • род 3: формировать объективную и аргументированную позицию по отдельным вопросам истории азербайджана.
   • • род 5: применять навыки сопоставления и установления взаимосвязей между древней историей азербайджана и современным периодом.
7. 7.
   Model evaluation and optimizationLecture

   The importance of evaluating machine learning models and the principles of correct determination of model performance are studied within the subject. the division of data into training and test (train/test) parts, as well as the concept of cross-validation and basic application principles are reviewed. approaches to evaluating, comparing the results of different models and selecting the appropriate model for a specific problem are investigated.

8. 8.
   Clustering methodsLecture

   Within the topic, the concept of clustering, its main purpose and its role in unsupervised machine learning are studied. principles of dividing data into groups based on similar characteristics and basic clustering approaches are reviewed. the working principles and applications of basic algorithms such as k-means, hierarchical clustering, and density-based clustering are explored.

9. 9.
   Artificial neural networksLecture

   Within the subject, the concept of artificial neural networks, their basic principles and their role in machine learning are studied. a network structure consisting of artificial neuron, input and output layers, hidden layers, weights and activation functions is considered. overfitting and underfitting problems associated with overfitting or underfitting the model to the data, as well as basic regularization methods for reducing these problems, are explored.

10. 10.
    Deep learning and CNNLecture

    Within the topic, the concept of deep learning, its main principles and role in machine learning are studied. the main advantages of deep learning models, such as the differences from traditional neural networks and the automatic learning of features from complex data, are reviewed. the structure of convolutional neural networks (cnn), basic functions and working principles of convolution, pooling and fully connected layers are investigated.

11. 11.
    Natural language processing (nlp)Lecture

    Within the scope of the subject, the concept, main purpose and fields of application of natural language processing (nlp) are studied. methods of text data processing, tokenization and segmentation of text into words and other units are considered. the application principles of machine learning and deep learning methods to NLP problems such as text classification, meaning and content analysis are explored.

12. 12.
    Generative artificial intelligenceLecture

    Within the framework of the topic, the concept of generative artificial intelligence, its main principles and role in machine learning are studied. the main features of generative models that allow the creation of new text, images and other types of content are reviewed. objectives, working principles and application features of discriminative and generative models are compared.

13. 13.
    Artificial intelligence agentsLecture

    Within the subject, the concept, basic principles and working mechanisms of artificial intelligence agents are studied. the principles of how agents perceive the environment, process the received information and perform appropriate actions to achieve certain goals are considered. basic agent types such as simple reflexive, model-based, goal-based, utility-based, and learning agent are explored. the role and interaction of the concepts of perception and action in the activity of agents is also explained.

14. 14.
    Ethical issues and social implicationsLecture

    Within the framework of the topic, the main ethical issues and social implications related to the application of artificial intelligence and machine learning technologies in cyber security are studied. ethical principles, data privacy, personal data protection and user rights are considered in cyber security activities. ethical risks related to data collection, processing and decision making of artificial intelligence systems are explored. it also explains the role of the main legal frameworks and legislative requirements applicable in the field of artificial intelligence and cyber security.

1. E. sorhun, "machine learning with python". 2021.
2. N. ismayilova, "artificial intelligence: machine learning methods". 2023.
3. Soma halder, hands-on machine learning for cybersecurity. 2018
4. Aurélien géron, "hands-on machine learning with scikit-learn, keras, and tensorflow." 2019