A.R. USIOBAIFO

Machine learning has become an important technology used in many areas such as healthcare, banking, education, and mobile applications. However, traditional machine learning methods usually require large amounts of data to be collected and stored in one central location, which can create privacy and security concerns. This project focuses on the design and implementation of a Federated Learning system for privacy-preserving machine learning. Federated Learning is a modern approach that allows multiple users or organizations to train a machine learning model together without sharing their raw data. Instead of sending personal or sensitive information to a central server, each participant trains the model on their local device and only shares model updates. This helps to protect privacy while still allowing the model to learn effectively. In this project, a federated learning prototype was developed using Python and machine learning tools. The system was designed with features such as secure aggregation and differential privacy to improve data protection and reduce the risk of information leakage. The performance of the system was tested using distributed datasets to simulate real-world situations where data is stored in different locations. The results showed that the proposed system was able to maintain a good level of model accuracy while protecting user privacy. It also demonstrated that federated learning can be a practical solution for organizations that need to use machine learning without exposing sensitive data. The study concludes that federated learning has great potential for institutions such as the University of Benin and other organizations in Nigeria that require secure and privacy-aware machine learning solutions.

A complete system design and implementation abstract framework tailored for your AI-driven spam email detection project is detailed below.AbstractElectronic mail remains a foundational pillar of global digital communication, yet its utility is continuously threatened by the exponential growth of unsolicited messages, phishing attempts, and malware-laden spam. Traditional rule-based and heuristic filters are increasingly obsolete due to their rigidity and inability to adapt to the highly evolving and sophisticated obfuscation tactics employed by modern spammers. To address these critical vulnerabilities, this project designs and implements an intelligent, adaptive Spam Email Detection System leveraging Artificial Intelligence (AI) and Natural Language Processing (NLP) techniques. The primary objective is to build a high-accuracy, real-time pipeline capable of automatically distinguishing legitimate messages ("ham") from malicious content ("spam") with minimal human intervention. The system's structural architecture comprises five core pipelines: data ingestion, text preprocessing, feature engineering, AI model classification, and deployment. During the design phase, raw email datasets (including text body, headers, and metadata) are subjected to rigorous NLP preprocessing, which includes tokenization, stop-word removal, lowercasing, and lemmatization to strip out textual noise. Feature extraction is then executed using TF-IDF (Term Frequency-Inverse Document Frequency) and Word2Vec embeddings to convert cleaned unstructured text into dense, high-dimensional numerical vectors. For the classification engine, a comparative implementation analysis is conducted using multiple Machine Learning (ML) and Deep Learning (DL) architectures, specifically Naive Bayes, Support Vector Machines (SVM), Random Forest, and Bidirectional Long Short-Term Memory (BiLSTM) networks. The implementation was developed using Python and integrated into a responsive web application dashboard via the Flask framework, allowing users to input raw text or connect live mailboxes for real-time scanning. Experimental evaluation of the models on standard benchmark datasets (such as the Enron and UCI Spam SMS/Email datasets) demonstrates that the BiLSTM deep learning approach, when paired with semantic word embeddings, yields the highest performance, achieving an accuracy rate exceeding 98.2%, exceptional precision, and a drastically reduced false-positive rate. The results prove that integrating AI-driven semantic understanding into email security infrastructure provides a scalable, highly adaptive defensive barrier capable of continuously learning from new spam patterns and significantly hardening enterprise cybersecurity posture.

This project aims to identify current and emerging phishing techniques and evaluate the level of awareness and preparedness among internet users, with the goal of proposing adaptive prevention measures. The methodology involved reviewing advanced global phishing tactics (including deepfake video phishing and QR code phishing) and conducting a primary survey among Nigerian youths. The study found that while exposure to phishing threats is high, the practical ability of users to correctly identify and prevent attacks is low, leaving them susceptible to scams like investment fraud and social media phishing. The project concludes that phishing is fundamentally a human-centric problem and proposes adaptive measures focused on integrating continuous education, practical simulation drills, and technological safeguards like 2FA to strengthen user vigilance.