Artificial Intelligence and Computer Networks

The course “Artificial Intelligence and Computer Networks” examines the role of AI in the analysis, optimization, and management of network infrastructures. It presents core machine learning techniques—from decision trees and clustering algorithms to neural networks and reinforcement learning—with applications in QoS, anomaly detection, load prediction, and energy management of nodes.

The course combines theory with practical Python/Colab labs, where students apply algorithms to network flow data and network topologies. Its goal is to develop an understanding of how AI-driven policies can be applied for intelligent and efficient network management.

Upon completion of the course, students will be able to:

• Understand how Artificial Intelligence is applied to optimize networks and edge/fog infrastructures.
• Analyze and preprocess network data for training Machine Learning algorithms.
• Apply classical algorithms (Decision Trees, K-NN, SVM, Clustering) in QoS and flow analysis scenarios.
• Use neural networks (CNN, RNN, GNN) for traffic load prediction and anomaly detection in networks.
• Implement reinforcement learning methods (Q-Learning, Deep Q-Learning) for routing and node energy management.
• Design intelligent and efficient strategies for managing modern network infrastructures.

1. Fundamentals of Artificial Intelligence (AI). Categories of intelligent agents and their applications in computer networks. Intelligent agents for packet routing in networks.
   • Introduction to Edge/Fog Computing: Definition of the environment where agents operate (Cloud-Fog-Edge Architecture).
   • The Edge device as an intelligent agent.
2. State-space search. Uninformed search techniques for pathfinding in networks.
   • Network Topology as a Graph: Modeling nodes (Edge nodes) and links as a state space.
   • Pathfinding: Finding paths in static networks for data transfer from IoT to Cloud.
3. Heuristic searches and A. Adversarial search and minimax algorithms. A algorithm for network routing.**
   • Routing Optimization: Introduction to the concept of cost.
   • Comparison of static heuristic methods with dynamic methods.
4. Constraint Satisfaction Problems (CSPs). Basic principles of real-world planning. CSPs for bandwidth allocation in networks.
   • Resource Management: Satisfying battery and computational power constraints.
   • QoS Constraints: How Throughput and Latency requirements translate into mathematical constraints for bandwidth allocation.
5. Logic and logical agents. First-order logic and knowledge modeling. Network troubleshooting with logical agents.
   • Fault Detection: Introduction to the problem of diagnosis.
   • Decision-Making Rules: Creating rules for network management.
6. Reinforcement learning and Q-learning. Applications of reinforcement learning in networks.
   • Q-Routing
   • Multi-Armed Bandits
7. Fundamentals of Machine Learning. Deep Neural Networks and their applications in networks.
   • Decision Trees: Entropy, Information Gain for Offloading decisions.
   • k-NN: QoS Classification and data normalization (Min-Max scaling).
   • SVM: Edge/Cloud class separation and SVR for latency prediction.
   • Neural Networks
8. AI applications in resource allocation.
   • Unsupervised Learning: Clustering nodes and packets for load balancing.
   • Resource Allocation: Using prediction models for dynamic resource allocation in Fog/Edge environments.
9. Federated Learning. AI-driven networks and applications in IoT and 5G.
10. Model training techniques on smart devices. Deployment and execution of AI on mobile devices and embedded systems.
11. Traffic data analysis using ML. Using Graph Neural Networks for network topology modeling.
    • Traffic Prediction: Using Regression for traffic load forecasting.
    • Data Analysis: Dataset preparation, cleaning, and management of QoS metrics.
12. AI-driven autonomous networks.
    • Synthesis: Combining Q-Routing (for routing), SVM (for anomaly detection), and FL (for learning) to create a fully autonomous network.
    • Self-organizing Networks: Application of unsupervised algorithms for topology self-organization.
13. Course review and recap, and research prospects.
    • Summary of comparative advantages.
    • Future trends (6G, Zero-touch networks).