Akash Rai
Research Subjects
Research Interests
Akash Rai has a total of 5 co-authors. Here are the top co-authors by number of shared publications:
- Arun Singh Yadav
Department of Computer Science, University of Lucknow, Lucknow 226007, Uttar Pradesh, India
1 shared publication - Jaydip Kumar
Collage of smart computing, COER University, Roorkee, Uttarakhand, India
1 shared publication - Rinki
IT Department, Netaji Subhas University of Technology, Delhi, Delhi, India
1 shared publication - Shubham Shukla
Computer Engineering and Applications, GLA University, Mathura, Uttar Pradesh, India
1 shared publication - Soumya Shrivastava
Department of Computer Science and Engineering, Indian Institute of Technology (Indian School of Mines), Dhanbad, Jharkhand, India
1 shared publication
Latest Publications
- Open Access
- 15 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-5
Optimized Machine Learning Framework for Phishing Detection: A Comprehensive Study
Phishing attacks exploit human trust using deceptive emails and URLs. This attack poses a major security concern nowadays, as its contribution to global data breaches was around 36% until 2024. This work presents a comparative analysis of various machine learning models and highlights the features of importance for in-time detection of phishing. To attain the objective of feature selection, this work integrates the Chi-Square Testing and Recursive Feature Elimination techniques. We tested six machine learning classifiers for this over two publicly available ISCXURL-2016 Dataset and a Phishing Email Dataset. The result of these approaches showcased the reduction in processing time by up to 65%, making it appropriate for real-time systems due to fast processing capability. Experimental results show that the Random Forest model achieved a peak accuracy of 99.57% for the email phishing, while XGBoost surpasses others by achieving 99.83% accuracy on URL dataset. These results show the importance of feature selection and ensemble methods in combating phishing and ultimately offer the use of the proposed framework in resource-limited situations, such as email clients and web browsers.
- 15 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-5
Open AccessOptimized Machine Learning Framework for Phishing Detection: A Comprehensive Study
Phishing attacks exploit human trust using deceptive emails and URLs. This attack poses a major security concern nowadays, as its contribution to global data breaches was around 36% until 2024. This work presents a comparative analysis of various machine learning models and highlights the features of importance for in-time detection of phishing. To attain the objective of feature selection, this work integrates the Chi-Square Testing and Recursive Feature Elimination techniques. We tested six machine learning classifiers for this over two publicly available ISCXURL-2016 Dataset and a Phishing Email Dataset. The result of these approaches showcased the reduction in processing time by up to 65%, making it appropriate for real-time systems due to fast processing capability. Experimental results show that the Random Forest model achieved a peak accuracy of 99.57% for the email phishing, while XGBoost surpasses others by achieving 99.83% accuracy on URL dataset. These results show the importance of feature selection and ensemble methods in combating phishing and ultimately offer the use of the proposed framework in resource-limited situations, such as email clients and web browsers.