Data Science2025

Developing Robust Neural Network Models for Network Intrusion Detection

Bachelor's thesis comparing standard and differentiable-logic-enhanced neural networks for network intrusion detection on real DDoS traffic (CSE-CIC-IDS2018).

Methodology

Trained a baseline neural network against a differentiable-logic variant, evaluating robustness against previously unseen attack patterns.

Findings

The differentiable-logic model achieved higher ROC-AUC and a substantially lower false-positive rate, showing logical constraints can improve both accuracy and reliability in cybersecurity systems.

Tools & Methods

Neural NetworksDifferentiable LogicCSE-CIC-IDS2018

Energy Consumption Analysis of Web Applications Using Bayesian Modeling Energy Demand Forecasting and Surplus Analysis Across Nordic Regions