Objectives and scope

[yeaight7]

This pull request adds a comprehensive introduction chapter to the thesis document, establishing the motivation, challenges, approach, contributions, and scope of the research on reinforcement learning for network intrusion detection. The new content provides essential context for the rest of the thesis and clarifies the experimental and methodological boundaries.

Key additions include:

Motivation and Problem Framing:

• Explains the limitations of traditional intrusion detection systems and motivates the use of machine learning, particularly reinforcement learning, for handling asymmetric risk in network security.
• Identifies methodological challenges in applying RL to network intrusion detection, including data quality, data leakage, and the need for fair baseline comparisons.

Proposed Approach and Contributions:

• Describes the offline RL formulation, canonical feature schema, use of a missingness mask, and binary action space for the agent, with a focus on reproducibility and leakage-aware evaluation.
• Outlines the multi-stage validation ladder and the same-protocol comparison with a Random Forest baseline as core contributions.

Scope, Limitations, and Document Structure:

• Clearly states the experimental and operational scope, emphasizing offline evaluation and the limitations of the prototype.
• Provides an overview of the thesis structure and what each subsequent chapter will cover. (Fd8ba3c0