| Pub. Date | : Dec, 2021 |
|---|---|
| Product Name | : The IUP Journal of Information Technology |
| Product Type | : Article |
| Product Code | : IJIT11221 |
| Author Name | : Agwi Uche Celestine*, Imhanlahimi R E** and Bernard Olorunfemi Paul*** |
| Availability | : YES |
| Subject/Domain | : Engineering |
| Download Format | : PDF Format |
| No. of Pages | : 16 |
Phishing is common with websites pertaining to scholarships, grants, recruitment, auction and online payment platforms. The attackers create fake websites that replicate the legitimate ones by cloning or by editing the html codes of the target legitimate websites. When victius seek resources available on these websites they are possibly redirected to the phishing websites to obtain their sensitive information either for financial benefits or to completely disrupt and halt the information processing system. To tackle such problems, an optimized adaptive machine learning system for detection and classification of phishing websites using feature-based phishing detection approach was implemented. A combination of Adaptive Neuro-Fuzzy Inference System (ANFIS) and Particle Swarm Optimization (PSO) algorithms coined as PSO-ANFIS technique was used. The PSO searches the massive online dataset and real-time network packets to extract features and identify optimal solution (feature selection) for training ANFIS, while Case Base Reasoning (CBR) approach was used to match new instances against instances in the knowledge base. The PSO-ANFIS model was tested using real-time network packet features and dataset available in University of California, Irvine (UCI) data repository. Performance was measured using Root Mean Square Error (RMSE), Precision, Recall and Accuracy and compared with Artificial Neural Network (ANN) and ANFIS. The result from PSO-ANFIS showed significant improvements in prediction capabilities compared to ANN and ANFIS without optimization algorithm. This shows that optimized machine learning such as PSO-ANFIS can address problems of bias, features selection and curse of dimensionality facing existing ML techniques.
Phishing remains a fundamental security issue in the cyberspace. It is a technique utilized by attackers to obtain sensitive information such as usernames and passwords, credit cards details, bank account information or other credentials from users for malicious purposes such as hacking, identity theft, or for their financial benefit by stealing money directly from bank accounts (Barraclough et al., 2014). Phishing activities evolve rapidly because legitimate html codes of websites can easily be copied and edited to possibly redirect victims to the phishing website. Attacks on Internet users through phishing have increased tremendously in recent times due to increase in deployment of Information and Communication Technologies (ICT) in all spheres of human interactions, especially in financial transactions. Phishing activities vary in size and dimensions.
Phishing, Particle Swarm Optimization (PSO), Artificial Neural Network (ANN), Adaptive Neuro-Fuzzy Inference System (ANFIS), Case Base Reasoning (CBR)