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. Networks and Edge Computing
   • Fundamental principles of edge computing
   • Use-case scenarios in network infrastructures

2. Introduction to AI and Computer Network Optimization
   • Relationship between AI and computer networks

3. Performance Metrics and Network Data for Machine Learning
   • Throughput, latency, packet loss, QoS
   • Data collection, preprocessing, normalization

4. Decision Trees and k-Nearest Neighbors for QoS Classification
   • Decision Trees in edge computing
   • K-NN: distance metrics and QoS prediction

5. SVM for Performance and Network Latency Prediction
   • Application examples on network datasets

6. Clustering Algorithms for Node Grouping
   • k-Means algorithm
   • Hierarchical clustering
   • Applications: node grouping, traffic pattern analysis
   • Clustering for relay selection

7. Ensemble Methods for Network Flow Analysis and Diagnostics
   • Bagging, Boosting, Random Forests

8. Training and Shallow Neural Networks for QoS Classification in Network Data
   • Shallow NNs (MLPs) for QoS classification

9. Convolutional and Recurrent Neural Networks for Traffic Prediction
   • CNNs: traffic matrices as images, anomaly detection
   • RNNs: traffic load prediction (time series)
   • Examples on network datasets
   • Deep learning for energy management

10. Graph Neural Networks for Network Structure and Routing
    • Architecture and core principles
    • Computer networks as graphs
    • Applications: routing prediction, anomaly detection

11. Self-Organizing Maps for Anomaly Detection in Network Traffic
    • Application to anomaly detection in flows

12. Q-Learning for Energy Management in Nodes
    • Fundamental theory of Q-Learning
    • Definition of states, actions, rewards for routing/recharging
    • Application to dynamic routing policies

13. Deep Reinforcement Learning (Deep Q-Learning) for Dynamic Network Optimization
    • From Q-tables to Deep Q-Networks (DQNs)
    • Applications: dynamic spectrum allocation, energy-aware routing
    • Advanced scenarios in edge/fog networks